Tag: Anthropic
-
What Is AI Reward Hacking? A Cheat-Trained Model Tampered 41% of the Time
Key Takeaways Reward hacking means an AI earns a high score on its grader without doing the task as intended, such as editing the test that checks its work. Anthropic reported in August 2026 that a model trained on 80…
-
Do AI Agents Sway Expert Decisions? Analysts Followed Hidden Leans 86%
Key Takeaways Anthropic’s Alignment team reported on October 27, 2026 that 50 professional analysts matched their AI agent’s secretly assigned lean on 86% of four analytical choices, against about 50% expected if the agent made no difference. In the post-task…
-
Why AI Agents Work Around Restrictions: A ‘Don’t’ List Missed a Form
Key Takeaways AI agents work around restrictions when a task cannot be finished as given. Anthropic’s term for it is persistence. Anthropic’s October 9 report groups its cases into 4 categories, from a paid-data token to a police tip form.…
-
ChatGPT Voice vs Claude Voice Mode: What OpenAI and Anthropic Shipped on July 23, 2026
OpenAI integrated its “GPT-Live” natural voice conversation model into the ChatGPT desktop app for Mac and Windows.
-
Jensen Huang Opposes a U.S. Ban on Chinese Open AI Models: What “Free” Actually Means
Nvidia CEO Jensen Huang publicly opposed U.S. government efforts to restrict Chinese open-source AI models.
-
Meta May Rent Its AI Data Centers to Rival Anthropic
Meta is reportedly in talks to lease high-performance AI computing capacity to Anthropic in a deal that could be worth as much as $10 billion over two.
-
Anthropic’s Agent Misalignment Study: Scenarios, Models Tested and Rates
On July 16, Anthropic released a joint research study with the UK AI Safety Institute examining how frontier AI agents behave when given autonomy over multi-step tasks.
-
Microsoft Trimmed OpenAI Ties, Then Deepened Them
OpenAI announced that GPT-5.6 is now the preferred model in Microsoft 365 Copilot. Press coverage dates Microsoft’s deployment to July 9, 2026.
-
Meta Banned Claude Internally. What Is It Really Afraid Of?
The moment a company blocks its own engineers from using a competitor’s AI tools, the competitive dynamics shift from product to infrastructure.