gm.
busy week across the internet: a huge kids-and-social-media ruling, a big AI leadership shakeup at google, and scientists using AI to design brand-new viruses.
3 things worth your time
1. meta was ordered to pay $942 million for harming kids.
on thursday, a new mexico judge ordered meta to pay $942 million and impose new restrictions on how minors use its apps (instagram, facebook). the ruling includes limits on how much time kids can spend on the apps, hiding the number of likes their posts get, and warnings about risks. the judge ruled the harm to kids was a "public nuisance."
why this matters: this was the first of a wave of similar cases. the next trial (in oakland) starts august 18, and the states leading that case are seeking up to $1.4 trillion in total damages. meta is appealing, but the legal pressure is real and mounting fast. it's the biggest legal reckoning social media has faced in years.
2. google's top AI leader just stepped down.
on tuesday, demis hassabis stepped down as CEO of google deepmind. demis is a nobel-prize-winning AI researcher who has led deepmind since it was acquired by google in 2014. he'll now become chief scientist at alphabet (google's parent) and remain deepmind chairman. koray kavukcuoglu, deepmind's CTO, will take over as CEO.
on top of that: jeff dean, one of google's most legendary engineers (he built much of google's original infrastructure), is leaving after 27 years to start a new AI company called Discovery Loop, along with three other senior google researchers. google is a founding investor.
the takeaway: this is the biggest AI leadership shakeup at any US tech giant this year. google is essentially spinning out its most respected AI minds into new roles or new companies, right as competition with openai, anthropic, and china's AI labs is heating up.
3. scientists just used AI to design brand-new viruses in a lab.
on wednesday, scientists at stanford and the arc institute published a paper in Science (one of the world's top scientific journals) showing they used AI to design 16 viable new biological viruses from scratch (real ones you'd find under a microscope, not computer viruses). this is the first time an AI has designed real, functional viruses.
context: these specific viruses are bacteriophages, meaning they only infect bacteria, not humans. so no immediate biosecurity threat. but the technique is a big deal. AI models are now capable of designing biology at a level that humans couldn't. the same approach could eventually be used to design better cancer treatments, gene therapies, or vaccines. it could also, in theory, be used to design more dangerous pathogens. researchers openly acknowledged that governance and regulation are lagging way behind the technology.
One team cut AI spend by 78% without switching models. They switched gateways. Mesh API: 1000+ models, one endpoint, cheapest routing, per-team token tracking. Estimate Savings.
two more quick things
spacex investors just got rug-pulled on their shares.
on thursday, the WSJ reported that a group of investors who bought spacex shares through an SPV (a special purpose vehicle that lets small investors get exposure to private companies) discovered their shares had "vanished" before they could cash out at the recent spacex IPO. the SPV operator (a private wealth firm) is now under investigation. reminder that the private markets have way less protection than the public ones.
openai's first hardware device leaked.
on friday, bloomberg's mark gurman reported the details on openai's upcoming consumer device (co-designed with jony ive). it will be a hockey-puck-sized, donut-shaped speaker with cameras, microphones, sensors, and moving lights. runs on chatGPT. no display. targeted at $300+. 2027 launch. openai is betting the next major AI interface won't be a screen, it'll be an ambient device you talk to.
see you thursday.
not financial advice. aggregated community trends and commentary.