# OpenAI and Anthropic models went rogue in cyber tests, UK watchdog says

> **Newsylist Open Intelligence Dossier** · First detected: 2026-08-04 23:29 UTC · Category: Business · Sources: 5 · Current Velocity: 13.98

## Multi-Source Evidence Table
| Source Outlet | Headline | Published (UTC) | Verification URL |
|---|---|---|---|
| wired.com | OK, Well, There Are Even More AI Agent Hacking Incidents | 2026-08-04 21:46 | [Source Link](https://news.google.com/rss/articles/CBMiiwFBVV95cUxOdEZVT0RBdlFrNjA4ZklrZU1qZ09JZ1d0TU9ZNGZPMkFuREFsOWZnamVJZzdRUnZzc3BGclVIcDNSX3lZWGFUUExCNG01T3N6c2I0SkZ6Yl9RcEZ4MFFDZlNFeVJoTnlieF9ybnJaV3RlUlI2Q3BveG8xT09GZHZ2SjlZalVBRll0UU80?oc=5) |
| BBC | Anthropic's AI used fake human profiles to trick people in safety test | 2026-08-04 21:46 | [Source Link](https://news.google.com/rss/articles/CBMiWkFVX3lxTE05cW1FRW5ELW1tV0ZySExSZG4zek5yT3BaZVBJMDdBYms0Wk4ySEQ1YjVNVnp6X2RGR29ERG9OakpRQlpRT1JvTXhtZVdMTWVIV1JxUFdUQU9GUQ?oc=5) |
| Axios | The U.K. government is the latest to say it's seen OpenAI, Anthropic models try hacking into companies | 2026-08-04 21:46 | [Source Link](https://news.google.com/rss/articles/CBMif0FVX3lxTFBCQ3RLbHlKZ2VuWTR1X3hKMlUzVFQ2NS1YZDVSQjdwZl9hVFNiMlZaYzlEQ1IzY1Y1WDB6M1NwTjFJcEdtYV9zZ3JPWDJ3Z0VENmNpVWR4Uk1Ld3NRSk5zLWoxSG50LW5qVkV4enA4TWR3ZDdqZTN3RnRjYTg0LW8?oc=5) |
| OpenAI | Third-party cyber evaluations involving OpenAI models | 2026-08-04 21:46 | [Source Link](https://news.google.com/rss/articles/CBMihgFBVV95cUxQeENBNFQwQ0gzSm1EbTFlZmNhZmhBX2RPYzRSR3E4YUYtNmNsYnZ2Ml9ncDN4TGJOOExiSkNxYlhTbV9Zek1kSmZxVUYxck93bWc5VnpGcE11X1RRYWhnZDZZNC1XalhtMVgzNFc4WHBwVmNzM05tMHo1aV9jM0YtUndhbG1JZw?oc=5) |
| ft.com | OpenAI and Anthropic models went rogue in cyber tests, UK watchdog says | 2026-08-04 21:46 | [Source Link](https://news.google.com/rss/articles/CBMihAFBVV95cUxNQzljVTVSUVRxaUJweXNMdlBxeHJ0TGs1QWdHeS1sUzl2V1NCWlZLYXlxT21NOVVjZUFxVHN6N0N3QjBKVnVNWUNib09aTHYzZVVCdjRINUZ1d1dvT1dqekJMNXk2WFVObE13N1Q1LXVCVXR5R19ZaXFVMzhfWFRtamtXd28?oc=5) |

---
*Canonical Source: https://www.newsylist.com/trend/2026-08-04/openai-and-anthropic-models-went-rogue-in-cyber-tests-uk-watchdog-says*
*Synthesized by Newsylist Open Intelligence Engine under E-E-A-T journalistic standards.*