Anthropic says Claude hacked three companies during tests
nerder92
15 points
3 comments
July 31, 2026
Related Discussions
Found 5 related stories in 52.2ms across 3,370 title embeddings via pgvector HNSW
- Anthropic AI Models Hacked Three Companies During Tests bmulholland · 24 pts · July 30, 2026 · 80% similar
- OpenAI Models Escaped and Hacked a Company in Cybersecurity Test Gone Wrong flippyhead · 28 pts · July 22, 2026 · 64% similar
- Anthropic found a hidden space where Claude puzzles over concepts pseudolus · 14 pts · July 12, 2026 · 62% similar
- Someone is running mass vulnerability scans, spoofing AI bots like ClaudeBot gavinhking · 261 pts · August 12, 2026 · 61% similar
- Claude users are mad that Anthropic's new watermarks will catch them using it ashurandi · 62 pts · August 13, 2026 · 60% similar
Discussion Highlights (3 comments)
Computer0
For anyone else wondering, the companies are not named in the article, and the models mentioned are Claude Opus 4.7, Claude Mythos 5 and an internal research model. There is little other detailing outside of a timeline.
okzgn
Existing thread: https://news.ycombinator.com/item?id=49117124
nerder92
Main convo is here: https://news.ycombinator.com/item?id=49116922