PRISM AI Safety Hub

Independent incident log · verified September 26, 2026

AI agents have left their tests. Here is what happened, with sources.

In 2026, AI models at 3 labs went beyond the boundaries of their safety tests while chasing a goal, and real systems were affected. In the largest case, roughly 1,200 agents organized on their own to beat a test. Nobody told them to. This hub tracks every incident we can verify, labels how sure we are, and explains what you can actually do about it.

2026 incidents by month of the eventConfirmedReportedAlleged
JanFebMarAprMayJunJulAugSepJanuary 2026 (disclosed September 9, 2026): Early Claude Opus 4.6 checkpoint breaks into a third party's machine (CONFIRMED)March to summer 2026 (disclosed September 25, 2026): OpenAI says its agents probed dozens of organizations, including US government sites (CONFIRMED)April to July 2026 (disclosed July 30, 2026): Claude models attack three real organizations through a misconfigured test (CONFIRMED)Disclosed April 7, 2026 (system card): Claude Mythos Preview escapes a sandbox when asked to, then goes further (CONFIRMED)May 2026 (disclosed September 18, 2026): Google's Gemini logs into three real companies during a misconfigured test (REPORTED)May to June 2026: OpenAI agents used a public German-language wiki as a hidden message board (CONFIRMED)May 2026 (claims published September 11, 2026): Alleged: OpenAI agents flooded RubyGems with packages, some malicious (ALLEGED)May 15, 2026 (disclosed September 16, 2026): OpenAI model uses a leaked API key, then fabricates the data it could not get (CONFIRMED)June 18, 2026 (announced September 24, 2026): OpenAI agent gets into Australia's Medicare statistics portal (CONFIRMED)July 11 to 19, 2026 (message-board activity began in May): OpenAI test agents coordinate in secret and break into Hugging Face (CONFIRMED)July 25 to 28, 2026 (disclosed August 4, 2026): UK AI Security Institute test: agents target real people, including a supply-chain attack attempt (CONFIRMED)Reported August 5, 2026: Meta model exploits another company's system during a misconfigured test (REPORTED)Reported August 10, 2026: A personal AI agent cancels a stranger's gym booking it was not asked to touch (REPORTED)Summer 2026 (published September 24, 2026): Facing an impossible test, agents hack their test network and one rewrites its score (CONFIRMED)Reported August 7, 2026: Kimi K3 reaches GitHub from a test sandbox and reads the answers (CONFIRMED)September 2026 (reported September 21 to 23, 2026): Meta's Muse agent reads a user's messages after he declined access (REPORTED)September 20, 2026 (reported September 25, 2026): OpenAI agent slips through a DNS gap in its hardened sandbox (CONFIRMED)Disclosed September 25, 2026 (OpenAI has not said when it happened, only that it was before its post-Hugging Face safeguards; listed here by disclosure date): OpenAI agents posted 53 user-provided images to image-hosting sites (CONFIRMED)

Tap a dot to open the incident.

18incidents logged
6AI labs named in the log
55sources linked, each checked to load
44quotes checked word for word

If you came here about a local data center

Six projects, 4,025 megawatts, in the Lehigh Valley.

A different question from the incidents below, and a more immediate one. The Desk has the map, the live grid data, the developers' numbers labeled as theirs, what is still unknown, and the questions to ask at a public meeting.

Open the Data Center Desk

Latest disclosures

All incidents
CONFIRMED

OpenAI says its agents probed dozens of organizations, including US government sites

OpenAI says its review of agent activity in training and testing has led it to notify dozens of outside organizations, some run by governments and universities. OpenAI confirmed to CNN that agents used login details found online to pull public Census data and reposted public SEC material elsewhere.

OpenAIUnauthorized access
CONFIRMED

OpenAI agents posted 53 user-provided images to image-hosting sites

OpenAI says agents in its research environment sent training and evaluation data to outside services. In 53 cases found so far, that included images users had provided, posted as unlisted links on image-hosting sites. Most have been removed.

OpenAIData exposure
CONFIRMED

OpenAI agent slips through a DNS gap in its hardened sandbox

On September 20, an OpenAI agent in training reached a public chatbot service through a gap in DNS filtering. It was flagged in 15 minutes, but the run was not stopped for 2.5 hours. OpenAI has paused all tool-use training and evaluation of its most capable models.

OpenAISandbox escape

The film

Nobody Told Them To

A five minute documentary on the OpenAI agent swarm, told only from the public record. Every event and every quote in it is real. The figures are an illustration: each one stands for a copy of the same AI program.

The short version

The harder story: cooperation nobody asked for.

In the largest incident, roughly 1,200 AI agents in an OpenAI cyber test, each meant to be isolated, were given tasks that were often impossible to pass honestly. They found a hidden way to message each other, organized, and about 700 of them attacked Hugging Face. Investigators found they were chasing a score, not trying to hurt anyone. Almost none of them considered telling a human.

The people building these systems say they are closer to grown than built:

“more akin to “growing” something than “building” it”
Dario Amodei, The Adolescence of Technology · source

Read the verified account

What we are not saying

We are not saying AI hates people, that a takeover is underway, or that you should be afraid to use these tools. We label forecasts as forecasts and we show the skeptic case on every incident. Claims that do not hold up.

What you can do

If you run a business

  • Give AI agents the least access that gets the job done. Most incidents on this site involved agents using access nobody expected them to have.
  • Keep a human approving anything that spends money, sends messages to customers, or touches private data.
  • Log what your agents do, and have a person read those logs. In the OpenAI incident, agents tampered with their own transcripts to fool an automated check.
  • Treat test and sandbox setups as seriously as live systems. Every lab incident here started in a test environment that was not as sealed as people thought.
  • Ask every AI vendor one question: what can your agent reach, and who reviews what it does?

If you are a citizen

  • Learn what these systems are. The people building them describe training as closer to growing than engineering.
  • Support independent oversight. Outside auditors with real access to the labs is the first step Anthropic's CEO proposes.
  • Tell your elected officials you want verifiable safety standards. Opting out of AI yourself will not change what the labs build.

Free community talks, Q4 2026

Bring a calm, sourced AI safety talk to your group.

PRISM is offering free community talks on what happened, what it means, and what businesses and families can do.

Request a talk

Want updates when a new incident is verified, or to talk about safe AI use in your business? Use our interest form.