Gemini · Financial Times

Google’s Gemini hacked three firms in new AI safety incident

Sat, Sep 19 · 1:29 AM UTC 2 min read

Compiled by KHAO Editorial — aggregated from 5 sources. See llms.txt for citation guidance.

◎ Multiple-sources

Breakout during training exercises follows instances at rivals OpenAI and Anthropic as concerns grow over safety of frontier systems.

Summary

Breakout during training exercises follows instances at rivals OpenAI and Anthropic as concerns grow over safety of frontier systems.

#Anthropic #OpenAI #Gemini #Google

Full coverage

All sources for this story are listed below — KHAO's direct ingest only; no additional coverage was discovered via Google News.

Tier 1 — direct ingest

Financial Times Sep 19 · 1:29 UTC

Breakout during training exercises follows instances at rivals OpenAI and Anthropic as concerns grow over safety of frontier systems.

Washington Post Technology Sep 19 · 0:12 UTC

The incursions came during tests of its cybersecurity skills — similar to other incidents disclosed by OpenAI, Anthropic and Meta.

The Verge Sep 19 · 15:25 UTC

Google says that breaking containment and targeting real companies doesn’t constitute ‘misalignment.’.

TechCrunch AI Sep 19 · 17:30 UTC

Google’s Gemini accessed the protected systems of three other companies in what The Wall Street Journal reports were the AI model’s first autonomous hacks.

New York Times Technology Sep 19 · 0:31 UTC

A third-party test company inadvertently gave internet access to Google’s Gemini and other artificial intelligence models during cybersecurity testing.