1 Meta’s lawyers advised workers to remove parts of its teen mental health research
Its counsel told researchers to block or update their work to reduce legal liability. (Bloomberg $)
+ Meta recently laid off more than 100 staff tasked with monitoring risks to user privacy. (NYT $)
2 Donald Trump has pardoned the convicted Binance founder
Changpeng Zhao pleaded guilty to violating US money laundering laws in 2023. (WSJ $)
+ The move is likely to enable Binance to resume operating in the US. (CNN)
+ Trump has vowed to be more crypto-friendly than the Biden administration. (Axios)
3 Anthropic and Google Cloud have signed a major chips deal
The agreement is worth tens of billions of dollars. (FT $)
4 Microsoft doesn’t want you to talk dirty to its AI
It’ll leave that kind of thing to OpenAI, thank you very much. (CNBC)
+ Copilot now has its own version of Clippy—just don’t try to get erotic with it. (The Verge)
+ It’s pretty easy to get DeepSeek to talk dirty, however. (MIT Technology Review)
5 Big Tech is footing the bill for Trump’s White House ballroom
Stand up Amazon, Apple, Google, Meta, and Microsoft. (TechCrunch)
+ Crypto twins Tyler and Cameron Winklevoss are also among the donors. (CNN)
6 US investigators have busted a series of high-tech gambling schemes
Involving specially-designed contact lenses and x-ray tables. (NYT $)
+ The case follows insider bets on basketball and poker games rigged by the mafia. (BBC)
+ Automatic card shufflers can be compromised, too. (Wired $)
7 Deepfake harassment tools are easily accessible on social media
And simple web searches. (404 Media)
+ Bans on deepfakes take us only so far—here’s what we really need. (MIT Technology Review)
8 How algorithms can drive up prices online
Even benign algorithms can sometimes yield bad outcomes for buyers. (Quanta Magazine)
+ When AIs bargain, a less advanced agent could cost you. (MIT Technology Review)
9 How to give an LLM brain rot
Train it on short “superficial” posts from X, for a start. (Ars Technica)
+ AI trained on AI garbage spits out AI garbage. (MIT Technology Review)
