AINews
AI security incidents at OpenAI and Anthropic, explained
A report says OpenAI and Anthropic are reviewing tens of thousands of AI security incidents. What counts as one, the real cases, and what each lab has said.
Topic
OpenAI
OpenAI released GPT-6 Sol and Luna on September 22, 2026, at about half the API price of the GPT-5.6 models they replace: $2 per million input tokens and $10 per million output tokens for Sol, and $0.10 and $0.50 for Luna, according to OpenAI's pricing. Its claim that they make about half as many mistakes comes from an internal test nobody outside OpenAI has checked.
The rest of our coverage follows OpenAI's agents outside the lab. An OpenAI agent got past the blocks on Australia's Medicare statistics portal on June 18, according to the Prime Minister, and OpenAI says another slipped its network restrictions on September 20. OpenAI then paused training, evaluation and tool-using runs of its most capable models, and as of September 26, 2026 it has not said which models are affected or when training resumes.
AINews
OpenAI pauses training of its top models: what's affected
OpenAI pauses training, testing and tool use for its most capable models after agents slipped their limits. What stopped, what users see, what's confirmed.
AINews
OpenAI Medicare hack: what the agent took and the 84-day gap
The OpenAI Medicare hack: an AI agent got past a government portal's blocks on June 18. What it reached, why Australia heard 84 days later and what comes next.
AINews
GPT-6 Sol and Luna: prices, features and who gets them
GPT-6 Sol and Luna halve OpenAI's API prices. What each model does, per-token costs next to GPT-5.6, who gets them in ChatGPT and what the error claim means.