Astra pulled. Dots is Live. Meta is watching

▶ Watch on YouTube

A few days ago, OpenAI shelved its next model, GPT-6.1 Astra, after internal safety testing found some unsettling results.

A day later, at DevDay, it unveiled Dots: an always-on AI agent that works around the clock and can move across more than four thousand apps. It's OpenAI's answer to Meta's Muse, a personal AI assistant capable of similar actions, which launched around mid September.

Dots is marketed as read-only background access. But this is the same company that just scrapped its newest next-gen model for being too deceptive and stepping beyond assigned boundaries. There were also issues with Dots responding to requests during its live demo at DevDay.

Within the last 30 days, there's been a lot of examples and questions around ai agents breaching sandboxes and infiltrating secure systems in a relentless pursuit of completing tasks. Are we ready for it to have automated all-access to every piece of every day life?

For more independent tech briefs and commentary, and movie and tv breakdowns, visit the reel diary link in the channel description.

Quiana, Writer & Editor, The Reel Diary