The real danger is not that machines will operate unsupervised; The point is that human observers have no idea what machines are actually “thinking.” Thankfully, science can suggest a way forward.
Read the full op-ed on the urgent need for new security measures around AI warfare.
Must read
I’ve scoured the internet to find you today’s funniest/important/scary/fascinating stories about technology.
1Despite Anthropic being blacklisted, the White House wants its new model
Trump officials are negotiating access to Mythos. (axios)
+ Anthropic said it was too dangerous for public release. (bloomberg $)
+ Finance Minister is concerned about security risks. (BBC)
+ anthropic A model has just been launched that is less risky than Mythos. (cnbc)
+ The Pentagon has waged a culture war against the company. (MIT Technology Review)
2 Sam Altman’s side hustle raises conflict-of-interest concerns
Their opaque investments could influence OpenAI’s decisions. (WSJ $)
+ A jury will soon decide whether OpenAI has abandoned its founding mission. (wired $)
+ The company is making a big play for science. (MIT Technology Review)
Starlink malfunction during 3 drone tests exposes Pentagon’s SpaceX dependency
This was one of several Navy testing disruptions involving Starlink. (reuters $)
+ DoD is also using Ford and GM for military innovations.(NYT $)
4 Data center delays threaten to block AI expansion
40% of this year’s projects are at risk of falling behind schedule. (foot $)
+Partly because no one wants a data center in their backyard. (MIT Technology Review)
5 Alibaba recently released its own version of the One World Model
Happy Oyster is the latest effort to enhance AI’s ability to understand physical reality. (SCMP)
+ But they still need to understand cause and effect. (foot $)
6 Google’s Gemini is now generating AI images tailored to personal data
By analyzing Google services and data of users. (quartz)
+ Google says this will reduce the need for detailed prompts. (techcrunch)
7 OpenAI is strengthening its agentive coding and development system
Its codec update is a direct attack on the cloud code. (The Verge)
+ But not everyone is confident about AI coding. (MIT Technology Review)
8 Europe’s online age verification app is here
It’s available for free to any company that wants it. (wired $)
9 Smartglasses are giving Korean theaters hope for a K-pop moment
Their AI-powered translations are taking the show around the world. (NYT $)
10 global voice actors fighting Hollywood’s AI push
Their voices are training the models who are taking their place. (rest of the world)
today’s thought
“There’s this dark period between now and some time in the future where the advantage is very aggressive AI.”
-Rob Joyce, former director of cybersecurity at the National Security Agency, explains bloomberg How AI is creating new hacking threats.
one more thing
