News Archive
Pick a date to read that day's stories.
Days with a blue dot have stories.
- 01 of 10OPENAI
OpenAI GPT-5.6 Goes Public After U.S. Safety Review
OpenAI's flagship Sol, Terra, and Luna models are now generally available after clearing a first-of-its-kind U.S. government frontier AI review.
OUR TAKEGovernment sign-off as a launch requirement means "shipped it" now comes with a security clearance badge.
Read more → - 02 of 10OPENAI
ChatGPT Work Agent Handles Tasks Across Apps for Hours
OpenAI launched ChatGPT Work, an agent that can act across apps, files, and workflows to complete multi-step projects.
OUR TAKEThe dream is an intern that never sleeps; the risk is a teammate that never stops Slacking you at 2 a.m.
Read more → - 03 of 10MICROSOFT
OpenAI Says GPT-5.6 Becomes Preferred Model in Microsoft 365 Copilot
GPT-5.6 will become the preferred model in Word, Excel, PowerPoint, and Cowork, right as reports say Microsoft is also building its own MAI models.
OUR TAKENothing says "we are still friends" quite like a press release during a public breakup watch.
Read more → - 04 of 10META
Meta Ships Muse Spark 1.1 for Agentic Coding
Meta released Muse Spark 1.1, a multimodal reasoning model with stronger tool use, coding, and computer-use capabilities.
OUR TAKEWhen OpenAI and Anthropic own the agent spotlight, a version bump with a dot-one is the loudest whistle you can blow.
Read more → - 05 of 10CHIPS
Meta's Custom AI Chips Enter Production in September
Meta plans to start producing its latest AI-specific chips in September to reduce dependence on scarce GPUs.
OUR TAKECustom silicon is the new designer handbag: everyone needs one, and the supply chain is the entire flex.
Read more → - 06 of 10ANTHROPIC
Anthropic's New Claude Feature Tries to Own Your Workflow
Anthropic rolled out a new Claude feature that proactively suggests and completes tasks across a user's apps.
OUR TAKE"Helpful" in the marketing deck is "intrusive" in the group chat.
Read more → - 07 of 10SPACEXAI
SpaceXAI Releases Grok 4.5, Its First Post-Cursor Coding Model
Elon Musk's renamed AI company launched Grok 4.5, the first model built after its Cursor partnership and aimed at coding and agentic work.
OUR TAKERebranding xAI to SpaceXAI and buying Cursor is the tech equivalent of a new haircut and a moving truck.
Read more → - 08 of 10CHINA
Beijing Considers Restricting Overseas Access to Top Chinese AI Models
Chinese authorities have held talks with tech firms about limiting foreign access to the country's most advanced AI models.
OUR TAKEExport controls are now a global two-way street, and the traffic light is red in both directions.
Read more → - 09 of 10REGULATION
Italy Fines Character.AI 158,000 Euros Over Age Verification Failures
Italian regulators fined Character.AI for failing to verify users' ages, exposing minors to potentially harmful chatbot interactions.
OUR TAKEA 158,000 euro fine is barely a rounding error for a frontier chatbot, but the paperwork is a warning shot for everyone else.
Read more → - 10 of 10HEALTH
Medical AI Looks Less Biased in Benchmarks Than in Real Care, Study Finds
A new Nature study found that large language models appear less biased in evaluations than they actually behave in clinical settings.
OUR TAKEThe gap between benchmark fairness and bedside fairness is where trust goes to get a second opinion.
Read more →