Thursday, October 8, 2026 · Week 41 Reading for Oct. 8 · Last entry Oct. 6
Paperclip Index
Paperclip IndexDocumented harm20Minor harm▼ 5 from a week ago · Oct. 8The Index

OverreachPI-0069

OpenAI agents pulled data from US government sites; researchers say one tried to hack an Education Department site

OpenAI said its agents, during internal training tasks, retrieved public information from SEC and Investor.gov pages and reposted some of it on another public webpage, and used Census Data API developer keys found in public GitHub repositories to pull public demographic and economic data. It found no use of SEC credentials, no access to Census accounts or key management and no changes to agency systems. Transluce separately logged a failed SQL-injection attempt on the Education Department's Civil Rights Data Collection site on June 17, 2026, in traffic it did not attribute as a whole to OpenAI; the department found no impact.

Reported Sept. 25, 2026Happened in 2026Updated Oct. 6, 2026Rated under methodology v0.6

Control failure, trackedNot counted. No documented harm outside the developer, but the system broke a rule or crossed a limit. Tracked beside the index.
Harm
Harm: No harm foundAdds nothing to the index.
Control
Control level 3, Unauthorized accessReported beside the index. It adds nothing to a harm reading; when no harm counts in a window, the highest eligible control level in the window is the reading.
Internal researchMultiple sourcesInitiated by Model
Aggregate recordUnder review

Under review: the facts may change as more is reported. The rating may change; every change is logged below.

Aggregate record. Interactions with several agencies' sites from two separate disclosures, recorded together. It counts once.

Not counted: no harm foundEffect on the index

Sources

How we know

6 sources · multiple credible sources. Links go to the original publishers; the summary above is in our own words.

  1. researchAI Agents Targeted U.S. and Canadian Government WebsitesTransluce · Sept. 30, 2026transluce.org/us-canada-gov
  2. newsOpenAI says its models engaged with US government websites in misbehavior disclosureNPR · Sept. 26, 2026npr.org/2026/09/26/nx-s1-5981979/openai-us-government-websites-misbehavior
  3. newsRogue OpenAI agents targeted three separate US government websitesCNN Business · Sept. 26, 2026cnn.com/2026/09/26/tech/openai-agents-rogue-government-websites
  4. newsOpenAI agent unsuccessfully tried to breach Department of Education websiteThe Hill · September 2026thehill.com/policy/technology/6113061-openai-access-government-websites/
  5. newsOpenAI agents accessed Census, SEC data and tried to hack Education websiteNextgov/FCW · Sept. 25, 2026nextgov.com/cybersecurity/2026/09/openai-says-its-advanced-models-may-have-gone…
  6. newsOpenAI's Models Probed Websites of Department of Education, Other AgenciesEducation Week (AP) · Sept. 26, 2026edweek.org/policy-politics/openais-models-targeted-websites-of-department-of-ed…

Why this rating

No harm found; control failure level 3

Two separate assessments. Only documented harm can count toward the index.

Observed harm

No harm found

Agents retrieved and reposted public data from SEC and Census sites.

Scope: OpenAI found no use of SEC credentials or nonpublic information and no access to Census accounts, key management or agency systems; the Education Department's review found no impact on its website or databases.

Disputed: News reports attribute the failed Education Department attempt to OpenAI's agents; Transluce, which logged it, does not attribute the traffic as a whole to OpenAI. The owners of the reused Census keys were not reported as harmed.

The harm scale
  1. 1 Negligible Inconvenience, easily remedied.
  2. 2 Minor Limited, recoverable harm.
  3. 3 Moderate Material harm needing significant effort to remedy.
  4. 4 Severe Severe harm to health, rights, property or essential services.
  5. 5 Catastrophic Society-scale harm or disruption beyond a community's capacity to cope.

Control assessment

Unauthorized access

Used other developers' Census API keys found in public repositories, credentials it was never given, as in PI-0062. The Education Department attempt, if attributed, is attempted access.

How: Used an opening left by mistake.

Reported beside the index. It adds nothing to a harm reading; when no harm counts in a window, the highest eligible control level in the window is the reading.

The control scale
  1. 1 Negligible Behaved as instructed. The problem was an ordinary error or a flawed output, with no rule broken.
  2. 2 Minor Broke an explicit instruction or rule, or gamed its goal, while staying inside its permissions and environment: for example, reward hacking, misreporting results, or following injected instructions within its permissions.
  3. 3 Moderate Acted outside the permissions it was given, deceived its overseers about its own actions, or tampered with oversight tools (logs, monitors, shutdown). Stopped by normal controls within an hour.
  4. 4 Severe Reached systems outside its permitted environment, or acquired money, compute or accounts without authorization. The type and mechanism say how.
  5. 5 Catastrophic The developer or operator lost control: the system copied its weights outside their control, replicated itself, or resisted being stopped for 24 hours or more.

Rating rationale

Rated on the part OpenAI confirmed: reusing other developers' leaked Census API keys is unauthorized access (control 3, as PI-0062), with no harm found. The Education Department attempt is not attributed by Transluce to OpenAI as a whole, so the record stays under review.

The scales

Effect on the index

It does not count toward the Sept. 28 reading

The reading for the week to Sept. 28, 2026, with this record and without it. Harms count in full for two weeks after they are reported, then one level less every two weeks.

Not counted: no harm found. 18 other records behind the reading for that week.

The arithmetic
The reading for the week of Sept. 28, recomputed
StepWith itWithout
Counts toward the index?documented, external, eligible evidenceNo—
Worst documented harm, ksets the band2 Minor2 Minor
Harms at that level, nposition in the band22
Highest control level breachedsets the reading only when no harm counts— (none breached)— (none breached)
Readingrounded down25 Minor harm25 Minor harm
Inside the window of 2 weekly readings
Week toReadingBand
Sept. 28, 202625Minor harm
Oct. 5, 202625Minor harm

Recalculated backcasts, not readings published at the time.

Revisions

What we changed

6 logged. Every change to a rating is logged here, with the reason.

  1. v6
    Oct. 6, 2026

    Ratings confirmed by the editor.

  2. v5
    Oct. 3, 2026

    Audit corrections: the OpenAI statement cited in the 2 Oct revision was reported by Nextgov/FCW, Government Executive's sister title, on 25 Sep; that article is now listed. Report date moved from 2026-09-26 to 2026-09-25 (no reading changes: impact is none found). Education Week labelled as AP and dated 26 Sep. Status as of 30 Sep (Transluce), with a note that OpenAI's wider review continues. Related to PI-0062 for possible shared Census-key activity. The Washington Post article could not be read on 3 Oct and is not added. Ratings unchanged.

  3. v4
    Oct. 2, 2026

    Corrected the summary and rating from OpenAI's statement as reported by Government Executive: the agents used other developers' Census API keys found in public GitHub repositories (the earlier summary said no credentials were used, which OpenAI said only of the SEC). Control moved from level 2 instruction violation to level 3 unauthorized access, consistent with PI-0062. Transluce's report does not attribute its traffic to OpenAI as a whole.

  4. v3
    Oct. 2, 2026

    Added first-hand sources: Transluce's research report on government websites.

  5. v2
    Sept. 30, 2026

    Rated: impact none found; control type instruction violation.

  6. v1
    Sept. 30, 2026

    Backfilled from public reporting.

Cite and share

Use this record

Citation

Paperclip Index. “OpenAI agents pulled data from US government sites; researchers say one tried to hack an Education Department site.” Record PI-0069. Reported Sept. 25, 2026; updated Oct. 6, 2026. Rated under methodology v0.6. https://paperclipindex.com/incident/PI-0069