The Grizzly Steppe report named Russian services and warned its IP list could also match legitimate traffic

On December 29, 2016, DHS and the FBI attributed intrusions to Russian intelligence services in a joint report released with new sanctions. Its indicators included computers used without their owners’ knowledge, and the report warned that matching traffic could be legitimate as well as malicious.

Looking Back revisits past weeks one at a time, up to the most recent, and asks what it meant for the banks, health systems and public agencies we serve. The series was written in 2026, with the benefit of hindsight, and each article says which week it looks back on.

Two-minute summaryThe dek and the first sentence of each section, quoted from the article

On December 29, 2016, DHS and the FBI attributed intrusions to Russian intelligence services in a joint report released with new sanctions. Its indicators included computers used without their owners’ knowledge, and the report warned that matching traffic could be legitimate as well as malicious.

  1. The week The report noted a change in practice.
  2. Attribution: naming the actors It also connected the report to an earlier statement.
  3. Indicators: leads that need review The fact sheet described the IP content.
  4. Tradecraft: behavior that outlasts indicators The report described how the two groups operated.
  5. Mitigations: controls that apply to any attacker Those questions are not specific to Russian intelligence services.
  6. How the report reached different organizations Financial institutions had two separate tasks.
  7. Objections, in order They are limited.
  8. What security leaders could check Start with mitigations.
  9. With hindsight The follow-up added detail where the December report had been thin.

The week

On December 29, 2016, the Department of Homeland Security and the Federal Bureau of Investigation released a Joint Analysis Report, reference number JAR-16-20296A, titled GRIZZLY STEPPE – Russian Malicious Cyber Activity. [1] It was marked TLP:WHITE, meaning it could be distributed without restriction. [1]

The report said it gave technical details on tools and infrastructure used by Russian civilian and military intelligence services to compromise networks and endpoints associated with the U.S. election and a range of government, political and private sector entities. [1] The government called this activity GRIZZLY STEPPE. [1]

The report noted a change in practice. It said previous Joint Analysis Reports had not attributed malicious cyber activity to specific countries or threat actors. [1] It said public attribution to Russian intelligence services was supported by technical indicators from the Intelligence Community, DHS, the FBI, the private sector and others. [1]

It described two actors. The report said one group, known as APT29, entered a U.S. political party’s systems in summer 2015, and a second, APT28, entered in spring 2016. [1] It said APT29 used a spearphishing campaign that reached over 1,000 recipients, including multiple U.S. government victims. [1]

The report said APT28 tricked recipients into changing their passwords through a fake webmail domain and used the harvested credentials to gain access. [1] It said actors likely associated with the intelligence services were still running spearphishing campaigns, including one days after the November 2016 election. [1]

The technical section pointed to indicator files in CSV and STIX formats and included one Yara signature. [1] Under actions to take, DHS recommended that administrators review the IP addresses, file hashes and signature, add the IPs to watchlists and review perimeter netflow or firewall logs. [1]

The report added a caution. It said organizations might find numerous instances of these IPs attempting to connect, and that some traffic might correspond to malicious activity and some to legitimate activity. [1] It said some traffic that looks legitimate, such as vulnerability scanning or browsing of public services, is in fact malicious. [1]

The report then gave mitigations. They included eight questions on backups, risk analysis, training, scanning and patching, application whitelisting, incident response, business continuity and penetration testing. [1] It listed seven strategies that DHS said can prevent as many as 85 percent of targeted cyber attacks. [1]

The report was released alongside other actions. A White House fact sheet the same day said the President had approved an amendment to the executive order on malicious cyber activity to cover interference with elections. [2] It said the President had sanctioned nine entities and individuals, including two Russian intelligence services. [2]

The fact sheet said the State Department had declared 35 Russian officials persona non grata, and that Russian access to two government-owned compounds would be denied. [2] It said DHS had added the report’s indicators to its Automated Indicator Sharing service. [2]

The amendment was Executive Order 13757, dated December 28, 2016, which took effect at 12:01 a.m. eastern standard time on December 29. [3] It amended Executive Order 13694 of April 1, 2015 to reach cyber-enabled activities that tamper with, alter or misappropriate information. [3] The new provision covered activities with the purpose or effect of interfering with or undermining election processes or institutions. [3]

The Grizzly Steppe report arrived on the same day as sanctions and diplomatic expulsions. [2] We read it as a document with four different kinds of content, each useful to a different reader and each with a different shelf life. The report itself mixed them in thirteen pages. [1] In our view, a security team that treated it as one thing would have used it less well than it could.

In our reading, the same mix recurs in later government reports; the February 2017 follow-up again combined signatures, tradecraft analysis and mitigations. [4] A report arrives, leaders want to know what it means, and the security team has to decide within hours what to load, what to search, what to change and what to tell whom. A clear split makes those decisions faster and easier to record.

Our argument is that a government threat report of this kind holds four separate products, and an organization gets value only when it sends each to the right owner. We take them in turn:

  1. Attribution: who did it, and what the government did about it.
  2. Indicators: IP addresses, file hashes and signatures to search for.
  3. Tradecraft: how the actors operated, described in words.
  4. Mitigations: the controls the report recommended, whoever the attacker.

The four parts are mutually exclusive. Attribution answers who. Indicators answer what to look for today. Tradecraft answers how the actors behave. Mitigations answer what to fix. Together they cover everything the report offered a network defender. After the four parts, we look at how the report reached banks, health systems and public agencies, take the strongest objections in order, and end with what security leaders could check.

Attribution: naming the actors

A change in practice

The report said previous Joint Analysis Reports had not attributed malicious cyber activity to specific countries or threat actors. [1] This one named Russian civilian and military intelligence services. [1] It tied the attribution to technical indicators from the Intelligence Community, DHS, the FBI, the private sector and other entities. [1]

It also connected the report to an earlier statement. It said its determination expanded on the Joint Statement on Election Security released October 7, 2016 by DHS and the Director of National Intelligence. [1] The attribution was a government position, released with sanctions the same day. [2]

Who uses attribution

For a private organization, attribution does not drive daily defenses. A bank’s security operations center blocks a malicious connection whoever made it. Attribution matters to other readers inside the organization. The board and senior leaders use it to judge whether the organization might be a target. Legal and compliance teams use it to check sanctions exposure. Communications teams use it to answer questions.

The report served those readers with a list of 48 entries under the heading of alternate names, mixing group names with malware and implant names. [1] That list let an organization map its existing threat reports, which used vendor names, to the government’s label. In our reading, that mapping was the attribution content a defender could use directly.

Sanctions as a separate track

The same day, the government took sanctions action. Executive Order 13757 amended Executive Order 13694 to cover persons responsible for cyber-enabled activities aimed at interfering with or undermining election processes or institutions. [3] It added an Annex listing persons whose property was blocked. [3]

The fact sheet said the President sanctioned nine entities and individuals, including the GRU and the FSB, four GRU officers and three companies said to have supported GRU operations. [2] It said Treasury was also designating two individuals under an existing part of the order for misappropriating funds and personal information. [2]

For financial institutions, those designations had a direct consequence. Blocked persons’ property in the United States, or in the possession or control of U.S. persons, could not be transferred, paid or otherwise dealt in. [3] That obligation ran through sanctions screening, not the security team. It is a separate use of the same week’s news.

Answering staff, customers and the board

Public attribution creates questions inside an organization. Staff ask whether their systems were involved. Customers ask whether their data is safe. Board members ask whether the organization is a target. Each question needs an answer that is accurate and does not overstate what the organization knows.

A useful answer has three parts: what the government said, what the organization checked in response, and what it found. The first comes from the report. [1] The second and third come from the organization’s own triage and control review. Without those records, the organization can repeat the news but cannot say anything about itself.

The diplomatic measures

The fact sheet also described diplomatic steps taken in response to what it called harassment of U.S. diplomatic personnel in Russia. [2] The State Department declared 35 Russian officials persona non grata and gave them and their families 72 hours to leave, and denied Russian access to two compounds in Maryland and New York. [2]

Those measures had no direct bearing on network defense. We mention them because they were part of the same announcement, and an organization’s leaders reading the news would have seen them together with the report. Separating the diplomatic, sanctions and technical content helps keep the security response focused.

What attribution does not change

Attribution does not change what a defender looks for. The tradecraft the report described, spearphishing and credential harvesting, is not unique to these two groups. [1] An organization that defends against it is defended against more than one adversary. The attribution told leaders who; it did not tell engineers what to build.

Indicators: leads that need review

What the report provided

The report pointed to indicator files in CSV and STIX formats and included a Yara signature named for a PHP web kit. [1] It told administrators to review the IP addresses, file hashes and signature, add the IPs to watchlists and look at perimeter netflow and firewall logs. [1] It encouraged them to check public-facing websites for the hashes and to run the signature on systems suspected of being targeted. [1]

The fact sheet described the IP content. It said the report included information on computers around the world that Russian intelligence services had co-opted without their owners’ knowledge, to make the activity hard to trace. [2] It said that in some cases the security community already knew this infrastructure, and in others the information was newly declassified. [2]

Why matches needed review

The report itself warned about matches. It said organizations might find numerous instances of these IPs attempting to connect, and that some traffic might be malicious and some legitimate. [1] That warning follows from the fact sheet’s description. A co-opted computer belongs to someone else, who may use it for ordinary purposes. Traffic from its address is not all hostile.

The February 2017 follow-up report made a related point about Tor. It said the majority of Tor traffic is not Grizzly Steppe activity, and that a Tor IP in a log only indicates that administrators should review the related traffic. [4] We use that later report here because it explains the reasoning behind the December warning. [1] [4]

Triage, not verdict

In our reading, the right use of the indicator list was triage. A match is a reason to look closer: at the direction of the traffic, the service involved, the time and what happened next. The report’s own examples point the same way. It said traffic that looks legitimate, such as vulnerability scanning or browsing public services, may be malicious, and that scans may precede exploitation attempts. [1]

An organization that blocks every listed IP without review may block legitimate users. An organization that treats every match as an intrusion may raise false alarms with leadership and regulators. The middle course is to log matches, review them against context and escalate only what the review supports.

The White House fact sheet leaned further toward blocking. It encouraged security companies and private sector owners and operators to use the indicators in proactive defense to block malicious activity before it occurs. [2] The report’s own advice for the IP addresses was a watchlist and a review of traffic. [1] An organization reading both documents had to reconcile them, and in our reading the report’s caution about legitimate traffic was the more specific instruction.

A hypothetical triage

A hypothetical regional bank loads the indicator list into its watchlist on December 30. The next morning it has several hundred matches, all inbound connections to its public website. The security team reviews them. Nearly all the connections, in this hypothetical, are ordinary page requests from addresses that also serve other users. A handful are scans of the login page.

The team records the scans, checks the web servers for the report’s file hashes and runs the signature on them. [1] It finds nothing. It writes a short note saying what it checked and what it found, and keeps the matches on watch. That note is what the bank can show its board and its examiners. The bank did not block the whole list, which would have cut off legitimate visitors.

The cost of false positives

False positives have costs beyond analyst time. An alert escalated to executives on a weak match can lead to a public statement or a regulatory notice that later has to be corrected. A blocked address can cut off a customer or a partner. Each of those costs grows when the match is treated as proof.

The report’s caution gave security teams cover to take the slower path. [1] A team that said a match was under review, rather than declaring an intrusion, was following the government’s own advice.

The fact sheet said DHS and the FBI were encouraging owners and operators to look back within their network traffic for signs of malicious activity. [2] That is a different use from blocking. It requires logs kept long enough to search.

For an organization with short log retention, the indicator list was of limited use for looking back. The report’s value in that case depended on decisions made before December 2016 about how long to keep netflow, firewall and DNS logs. That is a resilience decision with a cost, and the report made the trade-off visible.

Handling and onward sharing

The report was marked TLP:WHITE, which it explained meant the information could be distributed without restriction, subject to standard copyright rules. [1] Organizations could pass it to vendors, managed service providers and peers without asking permission.

That freedom matters for smaller organizations that rely on others for monitoring. A managed provider could apply the indicators across its clients the same day. It also matters for sector groups, which could discuss matches openly. Reports marked with more restrictive levels need a check before onward sharing.

Hashes and signatures

File hashes and Yara signatures behave differently from IPs. A hash match means a specific known file is present. A signature match means a file shares defined features with known malware. Both are stronger leads than an IP match, though a signature can still match benign files with similar structure.

The report’s signature was named PAS_TOOL_PHP_WEB_KIT. [1] DHS’s February 2017 analysis described the PHP files tied to the December report as web shells. [4] Its conditions included a narrow file size range and counts of specific strings, which made it precise. [1] In our reading, it was also easy for an actor to evade with small changes to the kit. A precise signature finds the known kit and little else, so it belongs beside broader web server monitoring rather than in place of it. That made it a test for internet-facing web servers, where such shells are installed after exploitation. The report tied the signature to its advice to check public websites and protect against SQL injection and cross-site scripting. [1]

Sharing channels

The fact sheet said DHS had added the indicators to its Automated Indicator Sharing service. [2] Organizations that received machine-readable feeds could load them without manual work. That speed is useful, but it carries the triage problem into automated systems. A feed that blocks automatically on every indicator applies the list without the review the report advised. [1]

How long indicators last

IP addresses change hands. Co-opted machines are cleaned or reassigned. Infrastructure is abandoned once exposed. The fact sheet anticipated part of this when it said defenders could use the malware information to force the intelligence services to re-engineer their malware. [2] Once that happens, the old hashes stop matching new samples.

In our reading, an indicator list is a snapshot. Its value is highest on release and falls as the actors adapt. An organization that loads it should also set a date to review and retire it, so that stale entries do not generate noise or block reused addresses.

Tradecraft: behavior that outlasts indicators

What the report described

The report described how the two groups operated. It said APT29 used spearphishing with links to a malicious dropper that delivered remote access tools and evaded detection. [1] It said APT28 used domains that mimic targeted organizations to trick victims into entering their credentials, and relied heavily on shortened URLs. [1]

It said both groups exfiltrated and analyzed information, used it to craft more targeted spearphishing, and set up infrastructure to hide their source. [1] It said APT29 used legitimate domains, including those of U.S. organizations and educational institutions, to host malware and send emails. [1]

The intrusion sequence

The report described the APT29 intrusion step by step. It said the actor delivered malware, established persistence, escalated privileges, enumerated Active Directory accounts and exfiltrated email from several accounts through encrypted connections. [1] For APT28, it described a spearphishing email that led recipients to change their passwords on a fake webmail domain. [1]

That sequence is a detection map. Each step leaves traces: a click on a link, a new process, a new persistence mechanism, a privilege change, directory enumeration, unusual outbound encrypted traffic. A security team can ask, for each step, whether it would see it.

The report said APT28 relied heavily on shortened URLs in its spearphishing emails. [1] A shortened link hides its destination until it is clicked. That defeats a user’s check of the domain, which the report’s phishing guidance otherwise relies on. [1]

Defenses include tools that expand and inspect shortened links before delivery or at the time of the click, and training that tells staff to treat shortened links in unexpected messages with suspicion. Neither depends on knowing the actor.

Credential harvesting

The APT28 description transfers readily to other threats. Credentials entered on a look-alike site let the actor log in as the user. [1] No malware is needed. Indicator lists do little against it, because the fake domain changes with each campaign.

The defenses are behavioral and structural: watching for logins from unusual places, requiring a second factor that a fake page cannot capture or replay, and teaching staff to check the domain. The report’s phishing guidance included checking URLs for spelling variations and using anti-phishing features. [1] The February 2017 follow-up recommended monitoring for traffic to sites that look similar to the organization’s own domains. [4]

Legitimate domains as hosts

The report said APT29 used legitimate domains, including some of U.S. organizations and educational institutions, to host malware and send spearphishing emails. [1] That weakens defenses that rely on domain reputation alone. A message from a respected university’s domain may still carry a malicious link if that domain has been compromised.

Defenses here look at content and behavior: links to unusual paths, attachments that run code, and messages that ask users to log in. They also include protecting the organization’s own domains so that they are not used against others, which is part of managing a web presence.

Exfiltration over encrypted channels

The report said APT29 exfiltrated email through encrypted connections back through its operational infrastructure. [1] Encryption hides the content. It does not hide the volume, timing or destination of traffic.

A security team can watch for unusual outbound volumes from mail servers and user endpoints, and for connections to destinations with no business purpose. Those signals do not depend on indicator lists. They depend on knowing what normal outbound traffic looks like, which takes baseline data the organization has to collect in advance.

Web servers as entry points

The report’s attention to web shells and injection flaws points to a second entry route. [1] It said connections from listed IPs might be vulnerability scans looking for cross-site scripting or SQL injection, and that exploitation attempts might follow. [1] The detailed mitigations included auditing external-facing systems, removing unauthorized code from web directories and validating input. [1]

For organizations with public websites, that guidance applies whatever the actor. A web server that accepts injection is a door for anyone.

Why tradecraft lasts

Actors change infrastructure faster than methods. The report described spearphishing by both groups, and credential harvesting through look-alike domains as an APT28 method. [1] In our reading, detections built on behavior keep working after the indicators expire. That makes the tradecraft section the part of the report with the longest useful life for a detection engineer.

Mitigations: controls that apply to any attacker

The questions

The report asked eight questions an organization might ask itself. [1] They covered offline backups and tested restoration, a risk analysis, staff training, scanning and patching, and application whitelisting. [1] The rest covered a practiced incident response plan, tested business continuity and penetration testing of its own systems. [1]

Those questions are not specific to Russian intelligence services. They are the basic program controls any security leader would recognize. The report presented them as a commitment to good practice. [1]

The seven strategies

The report listed seven mitigation strategies and said they can prevent as many as 85 percent of targeted cyber attacks. [1] They were patching applications and operating systems, application whitelisting, restricting administrative privileges, network segmentation, input validation, file reputation settings and understanding firewalls. [1] We cite the 85 percent figure as the report’s claim; we have not tested it.

The report said DHS continued to see intrusions because organizations failed to use these basic measures. [1] In our reading, that sentence is a useful line for a board. It says the defenses against a named state actor were the same ones the organization should already have.

Offline backups and tested restoration

The first of the report’s questions asked whether critical information is backed up, whether the backups are stored offline, and whether the organization has tested reverting to them during an incident. [1] Each part of that question is a separate control.

A backup that is online can be reached by an intruder with administrative access. A backup that has not been restored in a test may not restore when needed. The question asks for evidence of both. An organization that can show a dated restoration test answers it.

Privilege tiers and whitelisting

The strategy on administrative privileges said threat actors were increasingly focused on legitimate credentials, especially highly privileged accounts. [1] It advised reducing privileges to what each user needs and separating administrators into tiers with limited access to other tiers. [1]

That advice connects to the APT29 sequence, in which the actor escalated privileges and enumerated directory accounts. [1] Tiering limits how far an intruder can move with one stolen administrative credential. Application whitelisting, the second strategy, limits what code can run once the intruder is inside. [1]

Response and reporting

The report told organizations to implement their incident response and business continuity plans and to contact DHS or the FBI immediately if they found unauthorized access. [1] It gave contact routes for both. [1] For regulated organizations, that federal contact sits beside their own regulators’ notice rules, which counsel should check.

Spoofed mail

The phishing guidance began with a Sender Policy Framework record for the organization’s domain, to reduce the risk of receiving spoofed messages. [1] Such a record tells receiving servers which servers may send mail for the domain.

In our reading, the control works in two directions. It helps other organizations reject mail forged in the organization’s name, and it lets the organization’s own mail filters check incoming messages that claim to come from its domain. It is a low-cost step that a small organization can take without new tools.

Detailed guidance

The detailed mitigations covered SQL injection and web services, phishing, permissions and access controls, credentials and logging. [1] The credentials guidance called for a tiered administrative model with dedicated administrator workstations, to keep credential-theft tools such as Mimikatz from harvesting domain-level credentials. [1] The logging guidance asked organizations to turn on event logging, upgrade PowerShell for its enhanced logging and protect logs from modification. [1] The access guidance began with reducing privileges to those needed for each user’s duties. [1]

Why mitigations come first

Mitigations are the part of the report an organization can act on without any threat intelligence capability. They do not need a feed, a parser or an analyst. They need owners, budget and follow-through. For a small organization, they are the whole of what the report could offer.

How the report reached different organizations

Banks and other financial institutions

Financial institutions had two separate tasks. The security team could triage the indicators and review controls. The sanctions team had to screen for the persons blocked under the amended order. [3] The two tasks ran on different systems and different clocks, and an institution needed both.

Bank boards may also have asked whether their institution was a target. The report described targets across government, political organizations, think tanks, universities, critical infrastructure and corporations. [1] It did not single out banks. A calm answer to the board was that the controls the report recommended were the ones the bank should test regardless.

Health systems

Health systems were not named as targets in the report. [1] Their exposure came through the same methods: spearphishing, credential theft and internet-facing web applications. [1] A health system with a public patient portal or research website had reason to check the web server guidance and run the signature where appropriate.

Health systems with research programs also share traits with the universities the report named, including collaborators outside the organization and open networks for research. [1] Those programs are where the report’s guidance on legitimate domains used as hosts applies with particular force.

Federal agencies

Federal agencies were named among the targets. The report said the APT29 campaign reached multiple U.S. government victims. [1] Agencies had DHS as a direct partner and the indicators through federal channels. Their task was to search back through logs and confirm controls on the systems exposed to phishing.

Agencies also had to answer to oversight bodies. A record of what each agency searched, what it found and which mitigations it confirmed would let it answer questions from its leadership and from Congress without reconstructing events later.

State and local governments

State and local governments face the same methods with fewer staff. The report’s mitigations gave them a checklist that does not require threat intelligence tools. The intelligence community’s January 2017 assessment later said Russian intelligence had obtained and maintained access to elements of multiple state or local electoral boards. [5] That later finding gave election offices a specific reason to review the December report.

Universities and research institutions

The report named universities among the targets of both groups and said APT29 used educational institutions’ domains to host malware. [1] Universities therefore faced two risks: being targeted and having their own infrastructure used against others. Their open networks make the second risk harder to control.

Vendors and managed service providers

Providers that run systems for banks, health systems and agencies hold access that an intruder would value. The report’s advice on privileges, segmentation and phishing applies to them as it does to their clients. [1] A client can ask its providers what they did with the report, and expect a short written answer.

Smaller organizations

A small credit union, clinic or town may have no security operations center. For such an organization, the indicator files were hard to use. The mitigations and the reporting contacts were the usable parts. [1] A managed security provider could apply the indicators on the organization’s behalf, with the same triage caution.

Objections, in order

1. Indicator lists are useless

They are limited. The report itself said matches could be legitimate. [1] They still help with retrospective search and with web shell detection through hashes and signatures. [1] [2] Their value is highest at release and falls as actors adapt.

2. Attribution is politics, not security

Attribution served policy and sanctions on the same day. [2] [3] For defenders, it mainly helped map the government’s label to existing threat reports. [1] The report’s security value sat in its other three parts.

3. The mitigations were generic

They were general, and the report said so by calling them best practices. [1] Generic controls that an organization lacks are still gaps. The report’s statement that intrusions continued because basic measures were missing is the reason to check them. [1]

4. A non-government organization was not the target

The report described targets across sectors, including corporations and universities. [1] The methods described were not limited to political targets. Defending against them protects against other actors using the same methods.

5. Blocking all listed IPs is the safe choice

The report advised adding the IPs to watchlists and reviewing traffic; it did not tell administrators to block the list outright. [1] The fact sheet did encourage using the indicators to block malicious activity. [2] [1] Blocking co-opted machines may block their legitimate owners. A watchlist with review is closer to the report’s advice.

6. The follow-up report replaced the first one

The February 2017 report built on the first and said JAR-16-20296 remained a useful resource for understanding the actors’ methods. [4] It added signatures and kill-chain analysis. [4] The four-part reading applies to both.

7. Government reports arrive too late to help

Some content arrives late. The report said the intrusions it described began in 2015 and 2016. [1] Its indicators served chiefly for looking back, and its tradecraft and mitigations for looking forward. [1] [2] A report does not need to be early to show an organization where its own controls stand.

What security leaders could check

The four parts give a short set of checks for the next government report of this kind:

  • Send attribution and sanctions content to leadership, legal and sanctions screening, with a short note on what it means for the organization.
  • Load indicators into a watchlist with a review process, and set a date to retire them.
  • Search back through logs for indicator matches, and record the retention period that limited the search.
  • Map the described tradecraft to detections, step by step, and note the steps you would not see.
  • Check each listed mitigation against your controls, and assign owners for the gaps.
  • Record what was done with the report, by whom and when.

Where to start

Start with mitigations. They need no special tools and apply to any attacker. The report’s eight questions make a one-page review that a security leader can take to the executive team. [1]

Then look at credential harvesting. The APT28 method described in the report needs no malware and defeats indicator lists. [1] Multifactor authentication that resists phishing, and monitoring for unusual logins, address it directly.

Finally, set the log retention period on purpose. The fact sheet asked organizations to look back through their traffic. [2] An organization can only look back as far as it keeps logs. Deciding that period in advance, with its cost, is a resilience choice that every later report will test.

Measures worth keeping

A few measures show whether an organization uses government reports well. The time from a report’s release to indicators loaded with review. The share of matches reviewed and closed. The number of mitigation gaps found in each report review and the number closed by the next one. The log retention period available for retrospective search.

Those numbers come from records the security team already keeps. None of them requires a threat intelligence platform; a spreadsheet maintained after each review is enough to start. Reported to leadership after each major report, they show whether the organization learns from reports or only reads them.

Recording the response

Each report review should end in a short record. It names the report and its date, who reviewed it, and which indicators were loaded and when. It also says what the search found, which mitigations were checked and which gaps were assigned to owners. The record takes little time to write.

Its value appears later. After an incident, investigators, regulators and insurers may ask what the organization knew and did. A dated record answers that question with facts. Without one, the organization has to reconstruct its response from memory and scattered tickets, which is slower and less convincing.

What the checks will not do

These checks do not make an organization safe from a capable state actor. They make sure the organization used what the government gave it. That is the standard a board, a regulator or a court is likely to ask about after an incident.

With hindsight

DHS published a follow-up on February 10, 2017, Analysis Report AR-17-20045, Enhanced Analysis of GRIZZLY STEPPE Activity. [4] It organized the activity along the phases of the cyber kill chain and gave host and network signatures by actor and implant. [4] It said JAR-16-20296 remained a useful resource for understanding the two groups’ methods. [4]

The follow-up added detail where the December report had been thin. It listed email themes the actors used, such as fake fax notices and fake Outlook Web Access logins, and a set of vulnerabilities they had targeted. [4] It recommended monitoring for traffic to look-alike domains and said such domains were not included in the original report because they mimic victims’ own domains. [4]

It also restated the caution on indicators. It said the majority of Tor traffic is not Grizzly Steppe activity, and that a Tor address in a log only means administrators should review the related traffic. [4] It said DHS had analyzed web shells, executables and a malicious document attributed to the actors in an updated malware report. [4]

On January 6, 2017, the intelligence community released a declassified assessment of Russian activities aimed at the 2016 election. [5] Among its judgments, it said Russian intelligence obtained and maintained access to elements of multiple state or local electoral boards. [5] It said DHS assessed that the types of systems targeted or compromised were not involved in vote tallying. [5]

The sanctions authority outlasted the moment. The national emergency declared in Executive Order 13694 is continued by presidential notice. As checked on November 15, 2026, the latest continuation in our sources is the Notice of March 24, 2026, which continued the emergency beyond April 1, 2026 for one year. [7]

That notice listed the orders that took further steps under the emergency. [7] They were Executive Order 13757, Executive Order 13984 of January 19, 2021, Executive Order 14144 of January 16, 2025 (in relevant part), and Executive Order 14306 of June 6, 2025. [7] It listed Executive Order 14110 of October 30, 2023 in relevant part, and noted that Executive Order 14148 had revoked it. [7]

Executive Order 14306 changed the order’s reach. It amended subsections 1(a)(ii) and 1(a)(iii) of Executive Order 13694, as amended by Executive Order 13757 and later orders, to replace any person with any foreign person. [6] In our reading, sanctions under those provisions now apply only to foreign persons. Sanctions teams should confirm the current text and Treasury’s guidance with counsel.

What held up from December 2016 is the structure of the report. In our reading, attribution, indicators, tradecraft and mitigations remain the main kinds of content in government threat reports, and the February 2017 follow-up kept that split while adding detail. [1] [4] The caution about legitimate traffic held up as well, repeated by DHS in February 2017. [1] [4]

What changed is the legal frame around attribution. Executive Order 14306 changed who could be designated under subsections 1(a)(ii) and 1(a)(iii); we found nothing in it that removed the election provision Executive Order 13757 had added. [3] [6] In our reading, that provision now reaches only foreign persons. For defenders, the durable lesson came from the mitigations and tradecraft, which applied before the attribution and still apply now. [1]

Spotted an error? Tell us and we will check it and correct the article where needed.

  • Grizzly Steppe
  • threat intelligence
  • indicators of compromise
  • attribution
  • sanctions
  • spearphishing