It was a quieter week this week in the mainstream AI news, but Anthropic quietly went about shipping a huge volume of developer-facing features and news continued to seep out about OpenAI’s pivot. Oh, and Figure 03 turned up at the White House.
In Ethics news, David Sacks has departed as the White House’s AI Czar, there’s more follow up to Anthropic’s battle with the Pentagon, and Wikipedia cracks down on the use of AI in article writing.
Good interview with Sundar Pichai in Long Reads too.
Claude can now control your Mac
Anthropic have had one helluva week, shipping Channels, Dispatch, Computer Use, Auto Mode, and iMessage support in the last 7 days. Much of this is developer-facing, but I think allowing your AI Assistant to control your computer from your phone is interesting more broadly.
Much of this is inspired by the approach OpenClaw took, primarily to give it access to more of the things that you usually do on your computer, including all your data, as well as a dedicated compute environment.
Anthropic’s approach is a bit different, but it also presupposes that you have a computer to connect to in the first place. Many consumers don’t have a personal computer at home and do everything they need to do on their phone, tablet, or work computer, and because of this I don’t think this will be a big hit with consumers, who would want something much more native and integrated into their mobile device.
Interesting ideas though!
Anthropic | Simon Willison | X
OpenAI Pivots
It’s no secret to anyone paying attention that OpenAI is going through a bit of an identity crisis at the moment. Or if it is supposed to be a secret, then it’s not being kept very well!
It’s been widely reported that OpenAI are pivoting to focus less on consumers and more on Enterprise. I think this comes from two places - one a recognition of the success that Anthropic has been having in this space, and two a desperate need to grow revenues faster. Part of this pivot is to drop all their ‘side quests’ which is why they just gave up on Sora (despite having recently inked a $1bn deal with Disney), erotic mode (🤮) and pulling back from Instant Checkout which they had been testing in the US for the last 6 months.
It’s the second reason that I think is the big driver here, but also the most confusing. Testing a paid ads model in ChatGPT is obviously a revenue driver, but so was Instant Checkout (they we taking a 2-4% cut of every sale) which is why I’m not sure why they’re pulling back from it.
It makes sense for OpenAI to stop all the ‘side quests’ they were on - it was getting a little ridiculous how many things they were trying to do at once and was starting to feel a bit like a ‘shotgun strategy’.
Focus is good. Revenue is good. But it still doesn’t feel (at least from the outside) that they’ve quite settled on what their strategy should be going forwards.
Reuters | TechCrunch | The Verge | TechCrunch
Melania Trump and AI powered robot named ‘Figure 03’ open White House summit
Yep…. this happened. Memes followed.
AI Ethics News
David Sacks is done as AI czar - here’s what he’s doing instead
AI boom risks widening wealth divide, says BlackRock’s Larry Fink
Elizabeth Warren calls Pentagon’s decision to bar Anthropic ‘retaliation’
Anthropic wins injunction against Trump administration over Defense Department saga
Sam Altman-backed fusion startup Helion in talks to sell power to OpenAI
Anthropic left details of an unreleased model, an upcoming exclusive CEO event, in a public database
Long Reads
Fast Company - Google CEO Sundar Pichai’s plan to make Gemini the only AI that matters
Artificial Ignorance - The Bots Are Reading Along
“The future is already here, it’s just not evenly distributed.“
William Gibson