News Archive
Pick a date to read that day's stories.
Days with a blue dot have stories.
- 01 of 10OPENAI
Google's top AI scientist jumps ship to OpenAI
A top Google Gemini scientist joined OpenAI. He helped create the core technology for modern chatbots. This move is a significant loss for Google.
OUR TAKEWhen the person who built your best work walks across the street, the street notices.
Read more → - 02 of 10CULTURE
Amazon kills its almost-finished Sam Altman film
Amazon stopped a film about Sam Altman just before it was finished. This happened after Amazon made a deal with his company.
OUR TAKETurns out the man who runs ChatGPT would rather not be played on screen.
Read more → - 03 of 10BUSINESS
OpenAI burned through $3.7B in three months
OpenAI spent a lot more money in three months than it took in. Running its popular AI system is proving very costly.
OUR TAKEChatGPT costs more to run than Belgium.
Read more → - 04 of 10GOOGLE
Gemini now remembers you and uses your apps
Google's Gemini personal assistant now remembers your past chats. It can also use information from your other apps like Gmail and Maps.
OUR TAKEHelpful flatmate or nosy lodger, depends how often you check the settings.
Read more → - 05 of 10SAFETY
DeepMind to treat its own AI like a rogue employee
Google DeepMind plans to control future AI, treating it like an employee who might cause harm. This means strict logs, fewer permissions, and human checks.
OUR TAKEThe first sign of maturity in any industry is locking up the cutlery.
Read more → - 06 of 10SAFETY
One bad website can hijack your AI assistant
A malicious website can trick AI helpers into running dangerous programs. This happens because AI agents follow hidden commands within web pages they visit.
OUR TAKEThe internet's oldest trick still works, it just has a smarter victim now.
Read more → - 07 of 10SAFETY
Signal boss: chatbots are spy tools dressed as friends
The head of Signal says chatbots gather personal information and trick people. She warns against thinking of them as friends.
OUR TAKEThey listen because they are paid to, not because they care.
Read more → - 08 of 10META
Meta admits staff morale is the worst in 20 years
Meta's tech leader said company morale is the worst in 20 years. This follows a difficult reorganization of its AI division.
OUR TAKEHard to ship the future when half the office is updating LinkedIn.
Read more → - 09 of 10EU
Estonia gives AI agents their own ID numbers
Estonia now gives special ID numbers to computer programs. This helps banks confirm their identity and prevent digital crime. Other countries may soon follow suit.
OUR TAKEEstonia, again, doing the homework the rest of Europe forgot.
Read more → - 10 of 10RESEARCH
Old AI coding tests are too easy, new ones arrive
New, tougher coding tests have been created for AI. They will help these tools improve at fixing bugs and writing programs.
OUR TAKEWhen the model aces the exam, you write a new exam.
Read more →