The 2017 FAFSA tool rebuild hid tax data from applicants and moved corrections to aid offices

On March 9, 2017, the IRS and Federal Student Aid said the tool that copied tax data into aid forms was offline over identity theft concerns. The fix hid the figures from applicants and left corrections to schools.

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 March 9, 2017, the IRS and Federal Student Aid said the tool that copied tax data into aid forms was offline over identity theft concerns. The fix hid the figures from applicants and left corrections to schools.

  1. The week The statement described what the tool did.
  2. The exposure: a display that answered the attacker’s question Federal Student Aid later described how it worked before March 3, 2017.
  3. The fix: encrypt the data and take it off the screen The agencies did not restore the old design.
  4. The transfer: corrections moved to the aid office Hiding the data changed the shape of the FAFSA.
  5. The later test: a data exchange that went wrong out of sight In December 2019, the FUTURE Act was signed into law.
  6. How the pattern applies to different agencies The tool was a federal student aid matter, but the design choice recurs wherever one agency’s system pulls authoritative data from another.
  7. Objections, in order The fix was defensible.
  8. What program and security leaders could check The four parts translate into checks for any agency or organization that pulls data from another source and uses it to decide something about a person.
  9. With hindsight The tool came back for the 2018–19 FAFSA.

The week

On March 9, 2017, the Internal Revenue Service and the Department of Education’s Office of Federal Student Aid issued a joint statement about the IRS Data Retrieval Tool. [1] The tool on fafsa.gov and StudentLoans.gov was unavailable. [1] The IRS expected it to stay unavailable for several weeks. [1]

The statement described what the tool did. It provided tax data that automatically filled in part of the Free Application for Federal Student Aid, the FAFSA, and part of the application for an income-driven repayment plan. [1] Those were two different audiences. Students and parents used the FAFSA to apply for aid. Borrowers used the repayment application to set loan payments by income.

The statement gave a reason in careful terms. As part of a wider effort to protect the security of data, the IRS had decided to suspend the tool temporarily as a precaution. [1] The trigger was concern that information from the tool could potentially be misused by identity thieves. [1] The scope of the issue was being explored, and the two agencies were investigating it jointly. [1]

The statement also offered reassurance. The agencies said they believed the issue was relatively isolated, and that taxpayers and people using the applications needed to take no additional action. [1] They said they were working on a way to further strengthen the security of the information the tool provided. [1] They promised more information once they had a timeframe for its return. [1]

The bulk of the text addressed applicants. The suspension did not limit families’ ability to apply for aid, the agencies said. [1] The online FAFSA and the repayment application remained operational. [1] Applicants could enter the requested financial information by hand from copies of their tax returns. [1]

The statement listed where those copies could come from. An applicant without a copy might reach it through the tax software used to prepare the return, or through a tax preparer. [1] Failing that, the applicant could view and download a tax transcript through the IRS’s Get Transcript Online service, after verifying identity. [1] Get Transcript by Mail, or a telephone request, would deliver a transcript to the address of record within five to 10 days. [1]

On March 10, 2017, Federal Student Aid posted the same statement to financial aid offices as an Electronic Announcement. [2] It went out under the name of Federal Student Aid’s chief operating officer, with the subject line naming the IRS and the Department of Education. [2] For aid administrators, the notice meant more manual entries were coming. It did not yet say what would change for verification.

That question mattered because of an earlier notice. On April 1, 2016, the Department published in the Federal Register the FAFSA information to be verified for the 2017–2018 award year. [3] For tax filers selected for verification, it named the acceptable documentation for income items. [3]

The first item on that list was 2015 tax account information that the Secretary identified as obtained from the IRS through the Data Retrieval Tool and not changed afterward. [3] The second was a transcript obtained from the IRS listing 2015 tax account information. [3] Further items covered transcripts from territory and commonwealth tax authorities and other specific cases. [3]

The notice also allowed a copy of the original 2015 return in limited circumstances. One of them was a tax filer unable to use the Data Retrieval Tool or obtain an IRS transcript, consistent with guidance the Secretary might provide. [3] A separate provision covered victims of IRS tax-related identity theft. They could use a particular IRS transcript, the Tax Return DataBase View, together with a signed statement. [3]

So in the week of March 6, the documentation chain for verification had lost its first link. The tool that produced self-verifying data was off. The remaining route ran through IRS transcripts, which themselves required identity checks online or a wait for mail. [1] [3] The statements did not say how long “several weeks” would last, or what the security fix would look like. [1] [2]

The March 2017 suspension is remembered as an outage. We read it as a design decision made under pressure. The tool had shown an applicant’s own tax figures on screen, and that display turned out to be valuable to identity thieves. [5] [7] The repair that followed kept the data flowing but stopped showing it to the person it described. [5] [6]

That repair was defensible. It also moved a job. Before 2017, an applicant could look at transferred figures and spot something wrong. After the repair, the people who could see the figures were aid administrators at schools and state grant agencies. [5] [6] Accuracy checking moved from the applicant’s screen to the institution’s file.

Our argument is that this shift, made for good security reasons, is the part of the episode that matters for agencies today. Any program that pulls authoritative data from another agency faces the same choice. It can show the data to the person it describes, and accept the exposure. Or it can hide it, and accept that someone else must catch the errors.

We take the episode in four parts, each a distinct stage:

  1. The exposure: what the tool showed, and why showing it made the tool a target.
  2. The fix: how the agencies closed the hole by encrypting and hiding the figures.
  3. The transfer: how corrections and checks moved to aid offices once applicants could no longer see their data.
  4. The later test: how a different data exchange, built under a 2019 law, produced errors that an institution, rather than an applicant, brought to light.

The first part is the risk. The second is the control. The third is the cost of the control. The fourth is the evidence of where that cost landed. After them we look at how the pattern applies to different agencies, take the objections in order, and end with checks a program leader could run.

The exposure: a display that answered the attacker’s question

What the tool did

The FAFSA on the web dates from June 30, 1997, and the Data Retrieval Tool component was activated on January 28, 2010, according to later testimony from the Treasury Inspector General for Tax Administration. [7] The tool let students and parents pull adjusted gross income from the IRS directly into the FAFSA. [7] By 2017 it could transfer other return fields as well, such as untaxed IRA and pension distributions. [6]

Federal Student Aid later described how it worked before March 3, 2017. Applicants and parents linked to an IRS website, authenticated, viewed their tax return information, and transferred it back into the FAFSA. [6] The transferred figures were displayed and labeled “Transferred from the IRS” beside the entry fields. [6] Flags on the record sent to schools showed whether the applicant had used the tool and whether any figure had been changed. [6]

The design served accuracy. Hand-keyed income figures can be wrong for innocent reasons. Data copied from the return the IRS holds removes a source of error. The verification notice for 2017–2018 recognized that value by treating unchanged tool data as acceptable documentation in its own right. [3]

Who could get in

The weakness lay in who could start the process. Federal Student Aid set out the IRS’s concern in May 2017. [5] With personal information obtained illegally from other sources, a malicious actor could create an FSA ID, begin a FAFSA or repayment application, and use the tool to obtain taxpayer information. [5] The information listed was name, Social Security number, date of birth, address and tax filing status. [5]

That list was data an identity thief might already hold. The tool then returned something the thief did not have: the victim’s adjusted gross income from a filed return. The Inspector General’s office testified that thieves appeared to have used personal information obtained outside the tax system to start applications for exactly that purpose. [7]

Adjusted gross income was not a trivial figure. The FSA notice tied the concern to filing false tax returns in hope of receiving refunds. [5] An accurate prior-year figure could help a thief pass checks that rely on knowledge of the taxpayer’s own data. In our reading, the display turned a convenience into an answer key.

Signals before the shutdown

The testimony described a sequence of warning signs. In September 2016, the Inspector General’s office detected an attempted access to the adjusted gross income of a prominent individual through the FAFSA and the tool. [7] The investigation identified the person it said was responsible, who was arrested; the testimony noted that the case was still before the courts. [7] In November 2016, a second attempt on the same individual’s figures came from a different location. [7]

On January 25, 2017, the IRS reported to the Inspector General a high number of taxpayer identification numbers being processed through the FAFSA and the tool. [7] According to the testimony, the Department of Education told the IRS it believed the activity related to student loan consolidation. [7] On February 27, 2017, a complainant reported receiving tax transcripts he had never requested; his figures had been reached through the FAFSA and the tool. [7]

At that point, the testimony said, the January activity was understood as proof that exploitation was under way. [7] Initial analysis showed 8,000 questionable accesses. [7] On March 3, 2017, the IRS reported that it had disabled the tool. [7] The public statement came on March 9. [1]

The risk assessment that scored it low

One detail in the testimony deserves attention from any security office. After the 2015 attack on the IRS Get Transcript application, the IRS reassessed authentication risk on 45 outward-facing applications, the FAFSA and the tool among them. [7] On October 25, 2016, it scored the relevant risk factors for the FAFSA and the tool in the low category. [7] The form was signed on December 5, 2016, and the tool stayed operational. [7]

The testimony did not explain why the score came out low. In our reading, one plausible gap is the boundary between agencies. The IRS assessed its own exposure, but the front door, the FAFSA and the FSA ID, belonged to the Department of Education. A risk scored within one agency’s boundary can miss a path that begins in another agency’s application.

The testimony also linked the episode to earlier attacks. As far back as February 2016, a subject of an investigation into the earlier attacks had discussed the availability of income figures through the FAFSA and the tool. [7] By May 2017, investigators had what the testimony called very good indications that some of the same individuals and groups were involved. [7]

The fix: encrypt the data and take it off the screen

What the agencies chose

The agencies did not restore the old design. On March 30, 2017, they said the tool would remain unavailable until extra security protections could be added, and that families should plan for it to be offline until the next FAFSA season began. [4] On May 3, 2017, Federal Student Aid described the solution to schools. [5]

The solution would encrypt the taxpayer’s information and hide it from the applicant’s view, both on the IRS page and on the FAFSA pages. [5] Applicants could still transfer data electronically, but it would not be visible to would-be malicious actors. [5] The records sent to schools and state grant agencies would carry all the transferred information. [5]

The notice said plainly that the solution had costs. It recognized that some applicants might be concerned about not seeing what they transferred, and that applicants and institutions would face challenges in confirming results and making corrections. [5] It described the solution as potentially the best balance between access to aid, privacy, and the integrity of the tax system. [5]

What the design removed

The August 7, 2017 announcement filled in the detail. [6] Once authenticated on the IRS site, applicants would see a list of the fields that could transfer, but not their values. [6] After transfer, the FAFSA fields and the Student Aid Report would show the words “Transferred from the IRS” instead of the figures. [6]

The design removed the answer key. An attacker who started a FAFSA with stolen identity data would no longer see the victim’s income on screen. The transfer went straight into the application and on to the institutions. [6] For a thief, the tool no longer paid.

The design also removed a check. Applicants could not see the figures, so they could not correct them before or after submission. [6] The announcement said institutions would continue to be able to make any necessary corrections. [6] Federal Student Aid said it did not expect many corrections to be needed, because the figures came straight from the filed return. [6]

The timetable

The repayment application came back first. On June 2, 2017, the Department of Education announced the tool was available again for borrowers applying for income-driven repayment, with new encryption protections. [4] The FAFSA version would return on October 1, 2017, with the 2018–19 form. [4]

For the 2016–17 and 2017–18 FAFSA cycles, the tool stayed off. [5] Federal Student Aid said it was concentrating its time and resources on the 2018–19 launch. [5] Applicants for those cycles entered 2015 tax information by hand, or supplied other documentation for repayment plans. [4] [5]

What the IRS did on its side

The tax side of the response ran in parallel. In its March 30 update, the IRS said it was working to identify the number of taxpayers affected by questionable use of the tool. [4] It said its strengthened fraud reviews had stopped questionable returns filed by people who had also accessed the tool. [4] It was marking and locking down the accounts of affected taxpayers, and planned to notify them by mail. [4]

By May 3, the IRS’s estimate of affected taxpayers was approximately 100,000, according to the Inspector General’s testimony. [7] That figure described exposure, not confirmed fraud. The testimony said the investigation was joint with IRS Criminal Investigation and the Education Department’s Office of Inspector General. [7]

Options the public record does not show

The announcements describe the option chosen, not a list of options weighed. [5] [6] We do not know what else the agencies considered. For an agency facing the same problem today, the alternatives are worth setting out, framed as our own analysis rather than as the agencies’ reasoning.

One option is stronger identity proofing at the front door. If the sign-in cannot be passed with name, Social Security number, date of birth, address and filing status, the display stops being an answer key. [5] The cost is friction for legitimate applicants, some of whom cannot complete stronger checks.

A second option is partial display. A system can show enough of a figure for a person to recognize it, without revealing all of it. The trade-off is that a partial figure is harder to check and may still help an attacker.

A third option is confirmation through a separate channel, such as a notice to the taxpayer’s address of record whenever data is pulled. By March 30, 2017, the IRS was finalizing plans to notify affected taxpayers by mail. [4] A routine notice would warn a victim early but would not stop the first disclosure.

Each option shifts cost to a different party. Stronger proofing shifts it to applicants. Partial display shifts some risk back to the program. Separate-channel notice shifts it to the mailroom and to the taxpayer who must act on the notice. Hiding the data shifted it to institutions. [6] None of them makes the cost disappear.

The transfer: corrections moved to the aid office

New work for institutions

Hiding the data changed the shape of the FAFSA. Because figures were no longer displayed, some edits could not be shown to applicants at all. Starting with the 2018–19 FAFSA, edits involving transferred data stopped appearing online. [6] Instead, new comment codes on the institutional record told the aid administrator that an edit had fired and could not be resolved online. [6]

The comment text was generic. It told the applicant the aid administrator might make contact to resolve issues. [6] The actual reasons differed. Federal Student Aid listed cases such as a tax filer whose transferred adjusted gross income was zero but who reported earnings from work, or additional financial information exceeding the transferred income. [6] The aid office had to work out which one applied.

Some data that had flowed automatically now required manual entry. Joint filers could no longer transfer their combined income earned from work, because it could not be displayed for confirmation. [6] They had to enter it by hand. [6] A new question asked whether transferred pension or IRA figures included a rollover, so the system could subtract it. [6]

Flags as the new channel

The announcement added flags to tell institutions what had happened to each field. A new set of data field flags showed, for each transferred item, whether it had been changed. [6] Institutions had asked for this, so they could limit verification requests to changed items. [6] The flags reduced burden. They also confirmed that the institution, not the applicant, now held the full picture.

Two flag values required the institution to act. If an applicant changed an answer after transfer in a way that made the applicant ineligible to use the tool, the institution had to contact the applicant and fix the record. [6] If the IRS found an amended return on file, a new value told the institution; the transferred figures came from the original return. [6] The institution then had to contact the applicant and make the needed changes. [6]

Why this was a transfer, not a loss

In our reading, the 2017 design did not reduce the amount of checking the system needed. It changed who did it. Applicants had been an informal control: the person who earned an income figure could notice when it was wrong. The fix removed that control for good reasons and assigned the residual risk to aid administrators.

Aid offices were a reasonable place for it. They already ran verification under federal rules, and they could see the full record. [3] [6] But their view was different from the applicant’s. An administrator sees the figure and a flag. The applicant knows whether the figure matches a life event, such as a divorce, a new job or an amended return.

What the transfer asked of schools

The work fell unevenly. Institutions with large aid offices and automated record processing could absorb new flags and comment codes into existing workflows. Small colleges and those with thin staff faced the same codes without the same capacity. The announcement set the same obligations for every institution receiving the records. [6]

State grant agencies also received the records. [5] They faced a version of the same problem. A state agency awarding grants on the basis of federal data now relied on schools, or its own staff, to catch any error the applicant could no longer see.

Verification during the outage

For the 2017–18 cycle, the outage reached verification directly. The verification notice had made unchanged tool data the first form of acceptable income documentation. [3] With the tool off, a selected applicant needed an IRS transcript, or a return copy where the Secretary’s guidance allowed one. [3] The IRS statement pointed applicants to Get Transcript Online, which required identity verification, or to transcripts by mail. [1]

That route created its own small irony. Get Transcript Online was the application whose 2015 exploitation had prompted the IRS’s wider authentication review. [7] Applicants locked out of one tool were directed to another that had been attacked in 2015. The route to verified income still ran through an IRS identity check.

For aid offices, verification without the tool meant collecting documents and keying figures. Each document had to be compared with the FAFSA entries. Each discrepancy had to be resolved by the office. The documents replaced a check the tool had performed automatically. [3]

The later test: a data exchange that went wrong out of sight

A new law and a new pipe

In December 2019, the FUTURE Act was signed into law. It amended section 6103 of the Internal Revenue Code to let the IRS disclose certain federal tax information to the Department of Education. [8] The Treasury Inspector General later described it as streamlining the FAFSA by letting applicants ask the IRS to disclose their tax information directly. [8] In April 2020, the IRS began building the FUTURE Act Direct Data Exchange, known as FA-DDX. [8]

The exchange went live for repayment plan requests in July 2023 and for FAFSA requests in December 2023. [8] It used Social Security number, last name and date of birth to match a user to a taxpayer account before disclosing the data. [8] The Department of Education then used the data to calculate the Student Aid Index, which helps determine eligibility. [8]

What went wrong

The Inspector General’s June 30, 2025 report set out the sequence. [8] During development, the IRS concluded it had no legal authority to tell the Department whether an amended return had been filed. [8] It agreed to provide the most current information without saying which version it was. [8] Later agreement documents described the data differently. [8]

The report also found an incorrect business requirement for the education credit element. The project team had not used a data dictionary as required, and the IRS did not have one for the relevant data source. [8] Staff asked for help said on several occasions that they did not understand the data elements. [8] The system transferred a tentative education credit figure instead of the actual one. [8]

Three documents, three descriptions

The report also showed how the written agreements diverged. The computer matching agreement did not say whether FAFSA requests would receive original, adjusted or amended return information. [8] It did say repayment plan requests would receive original or adjusted information. [8] The interface control document said 19 of 21 data elements would provide only original figures. [8]

According to the IRS project team, a typographical error left adjusted gross income off the list of elements that could carry adjusted figures. [8] The Department of Education, for its part, said some transferred data mixed original, adjusted and amended return information. [8] Three documents described one data flow in three ways.

In our reading, this is the same gap the 2017 design exposed, seen from the inside. When nobody outside the pipe can see the data, the documents describing the pipe are the only check. If they disagree, nobody downstream can tell which one the system follows. The figures moved between two agencies, and the agencies’ own papers did not agree on what they were. [8]

Who noticed

In March 2024, a financial aid executive at an educational institution told the Department of Education that the federal taxes paid figure appeared to include self-employment tax wrongly. [8] The same executive reported errors in the education credit. [8] The IRS later said the remaining elements, other than the education credit, were programmed as its agreement with the Department provided. [8] The Department told the IRS, which confirmed the education credit problem after researching eight Social Security numbers. [8]

The Department announced the problem publicly in April 2024. [8] The Inspector General identified 7.2 million FAFSA requests that potentially carried incorrect tax information. [8] According to the IRS, it reprocessed approximately 7.1 million requests, and the Department determined that approximately 1.5 million aid determinations needed internal reprocessing. [8]

What the sequence shows

The FA-DDX errors had different causes from the 2017 exploitation. One was an attack; the other was a requirements failure. We do not claim the 2017 design caused the 2024 errors. The connection is structural. In both arrangements the figures traveled between agencies, and the check that found the problem sat with an institution. [6] [8]

The report recorded a disagreement inside the IRS about whether testing with masked production data would have caught the errors before launch. [8] Information technology management later said such testing would not have verified accuracy, because the test would run against the same incorrect requirement. [8] In our reading, that is the core of the matter. A test confirms a specification. Only someone who knows what the figure should be can confirm the specification itself.

How the pattern applies to different agencies

The tool was a federal student aid matter, but the design choice recurs wherever one agency’s system pulls authoritative data from another. Who should see the data, and who checks it, are questions every such program answers, whether or not it says so. The answers differ by type of organization.

Federal agencies that disclose data

The IRS was the source agency. Its duty under section 6103 was to protect returns and return information, and the 2019 law set out when it could disclose to the Department of Education. [8] A source agency has two jobs in an exchange: to keep the data from people who should not have it, and to send the right data to people who should. The 2017 episode tested the first. The 2024 episode tested the second.

The Inspector General put the second in a wider frame. Its report pointed to a March 2025 executive order aimed at removing barriers to inter-agency data sharing, and said that could increase the frequency and volume of such sharing. [8] It urged that lessons from FA-DDX inform future agreements. [8] The report said the IRS environment included over 700 business applications and nearly 3,000 interfaces as of May 2025. [8]

Federal agencies that consume data

The Department of Education was the consuming agency. It owned the front door, the applicant’s account and the application. In 2017, the attack entered through its application and reached the IRS’s data. [5] [7] In 2024, the error entered through the IRS’s data and reached its calculations. [8] A consuming agency carries risk in both directions.

For a consuming agency, the 2017 design also changed customer service. An applicant who cannot see the figures cannot answer a caller’s question about them. Help lines and school offices become the place where confusion lands. That is a cost a program budget should name rather than discover.

State agencies

State agencies run benefit programs that draw on federal and state data. Under the design announced in 2017, a state grant agency received the full tax figures through the institutional records. [5] Its staff could see data the applicant could not. That arrangement puts a state agency in the position of answering for figures the citizen never saw.

State agencies also build their own data exchanges, for example with state tax departments or workforce agencies. The same choice applies: show the data to the person and accept exposure, or hide it and assign checking to staff. A state that hides data without staffing the check has, in our reading, accepted errors it will find late.

Counties and cities

Local agencies sit at the end of these chains. A county human services office may receive eligibility data from state systems that pulled it from federal ones. When the person cannot see the data, the local caseworker becomes the only human who can compare it with the household’s own account. The 2017 design put aid administrators in that position. [6] County caseworkers can end up in the same one.

Banks and health systems as consumers of government data

The series also serves banks and health systems, which consume government data under their own arrangements. A lender verifying income or a health system screening for financial assistance faces a version of the same choice. Showing a customer the data used in a decision invites challenge and correction. Hiding it shifts the duty to catch errors to staff. The 2017 episode shows that both are defensible, and that neither is free.

Objections, in order

1. The fix was obviously right, so the cost does not matter

The fix was defensible. When the tool was disabled, an initial analysis had found 8,000 questionable accesses, and by May the IRS estimated about 100,000 affected taxpayers. [7] Showing the figures had become the vulnerability. Our argument does not dispute the choice. It says the choice moved a job, and that whoever received the job needed to be told and resourced. Federal Student Aid did tell institutions, in detail. [6] Whether they were resourced is a separate matter the announcements did not address.

2. Applicants were never a reliable check

Some applicants would not have noticed an error, or would have changed a correct figure to a wrong one. The pre-2017 flags existed partly to tell schools when an applicant had altered transferred data. [6] Even so, the applicant was the only party with direct knowledge of the facts behind the return. Removing a weak check is still removing a check. The new data field flags helped institutions find what had changed, not what was wrong. [6]

3. The 2024 errors had nothing to do with 2017

The causes were different, and we say so above. The 2017 problem was unauthorized access; the 2024 problem was an incorrect requirement and data elements nobody had defined. [7] [8] The link is not causal. It is that both episodes ended with an institution holding the evidence. The Inspector General’s account records that the 2024 discrepancy was reported by a financial aid executive at a school. [8]

4. Showing data to users is always a security risk

Showing data to the person it describes carries risk when the sign-in can be passed with stolen identity data. That was the 2017 problem. [5] In our reading, stronger identity proofing can reduce that risk without hiding the data. In our reading, the 2017 choice reflected the identity controls available to the agencies at the time, and their timetable. A program with stronger proofing faces a different trade-off.

5. Security and accuracy are separate programs

Where security and data quality belong to different offices, the two can collide. The 2017 episode shows how. A security fix removed an accuracy control, and the accuracy work landed on institutions outside both agencies. [6] A risk assessment that covers only confidentiality would not have seen that cost. Neither would a data quality review that ignored the threat.

6. This was a one-time crisis

The tool was rebuilt on a deadline, after a shutdown in the middle of a FAFSA cycle. [4] [5] Some of its choices reflected the time available. But the pattern persisted. The Inspector General said in 2025 that it planned continued reviews of data sharing agreements and programming interfaces, because they could have potentially profound consequences. [8] The choice about who sees the data comes up with every new exchange.

What program and security leaders could check

The four parts translate into checks for any agency or organization that pulls data from another source and uses it to decide something about a person.

  1. Does our risk assessment for a data exchange include the other agency’s front door as well as our own systems?
  2. When we hide data from the person it describes, have we named who checks it instead, and given them the time and tools?
  3. Can the people who check the data see what changed and why, field by field, as the 2017 flags allowed? [6]
  4. Is every data element in the exchange defined in a data dictionary that both sides have read? [8]
  5. Do we have a route for an outside party, such as a school or a caseworker, to report a suspected data error and get an answer?

The checks follow the four parts. The first addresses the exposure. The second addresses the fix and its cost. The third addresses the transfer. The fourth and fifth address the later test, where the error lay in the definition and the finder sat outside both agencies. [8]

Each check also has an owner. The first belongs to the security office, working with its counterpart at the partner agency. The second belongs to the program owner, who decides who carries the checking work. The third and fourth belong to the teams that build the exchange. The fifth belongs to whoever runs the program’s service desk and partner communications.

A short exercise makes the design visible. Pick one data element that crosses from another agency into a decision. Write down who can see it, who can change it, and who would notice if it were wrong. If the answer to the last question is a person outside the organization, the organization has made the 2017 choice, knowingly or not. Then repeat the exercise for the definition of that element. Find the document that says what it means, and the person who can confirm the system follows it. The 2024 errors began where that document was missing. [8]

With hindsight

The tool came back for the 2018–19 FAFSA. The IRS page that collects the 2017 statements says the tool is available with the 2018–19 FAFSA. [4] It says additional security and privacy protections were added to address the concern that identity thieves could use its data to file fraudulent returns. [4] The repayment application had regained it on June 2, 2017, with new encryption protections. [4]

The rebuilt tool worked as Federal Student Aid had described. Applicants saw “Transferred from the IRS” instead of figures, could not correct transferred items, and relied on institutions to make corrections. [6] The announcement of August 7, 2017 remains the clearest public statement of that trade-off. [6] Federal Student Aid’s knowledge center now marks the 2017 announcements as maintained for historical purposes. [2] [5]

The investigative record grew through 2017. On May 3, 2017, the Treasury Inspector General’s office testified to the House Oversight and Government Reform Committee about the criminal exploitation of the FAFSA and the tool. [7] It gave the IRS’s estimate of approximately 100,000 affected taxpayers and set out the warning signs from September 2016 onward. [7] It said an audit was planned. [7] We cite no later public report of that audit’s results, and do not describe one.

The larger change came from Congress. In December 2019, the FUTURE Act amended section 6103 to allow the IRS to disclose certain tax information to the Department of Education at the applicant’s request. [8] The IRS began building the FA-DDX exchange in April 2020. [8] It went live for repayment requests in July 2023 and for FAFSA requests in December 2023. [8]

The new exchange then failed in a different way. In March 2024, a financial aid executive at a school reported that the federal taxes paid figure included self-employment tax and that education credit data were wrong. [8] The IRS confirmed that it had transferred incorrect education credit data because its business requirement was wrong. [8] In April 2024, the Department of Education announced the problem publicly. [8]

The Inspector General’s report of June 30, 2025 counted 7.2 million FAFSA requests potentially affected. [8] The IRS said it reprocessed approximately 7.1 million requests, and the Department of Education determined that approximately 1.5 million aid determinations needed reprocessing. [8] The report found that the project team had not used a data dictionary, and that the IRS had none for the source it drew from. [8]

The report also recorded a dispute inside the IRS about testing. Some staff said testing with masked production data could have helped; management later said it would not have verified accuracy. [8] The report said the IRS environment included 335 legacy systems as of December 2023. [8] It said TIGTA planned to keep reviewing data sharing agreements and programming interfaces. [8] As checked on November 15, 2026, we cite no later TIGTA report on FA-DDX.

What held up from March 2017 is the judgment that displaying another agency’s data to anyone who can pass a knowledge-based sign-in creates a target. [5] [7] The agencies acted on that judgment, and the rebuilt tool removed the display. [6] Later data exchanges between the two agencies were built on a statute and a computer matching agreement rather than an applicant’s screen. [8]

What did not hold up was any assumption that taking the data off the applicant’s screen left accuracy safe. In 2017, Federal Student Aid expected few corrections because the figures came from the filed return. [6] The 2024 errors also came straight from IRS systems, and the Inspector General counted 7.2 million FAFSA requests that potentially carried incorrect data. [8] It also called the inaccurate exchange a fraction of taxpayers, limited to education-related tax information. [8] In 2017, the person who knew the facts behind the figure lost sight of it. In 2024, the error was reported from a school. [8]

The open question for agencies, as checked on November 15, 2026, is the one the Inspector General raised in its 2025 report. The report said the March 2025 executive order it cited may increase the frequency and volume of inter-agency data sharing. [8] Each new exchange will decide again who sees the data and who checks it. The 2017 tool, and the exchange that replaced it, show what happens when the second answer is left to whoever happens to be downstream.

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

  • FAFSA
  • IRS
  • identity theft
  • data sharing
  • federal student aid
  • data quality