AI news briefs, Sept. 29 — Oct. 4, 2025

From “AI-designed toxins slip through safety checks” through “Sam Altman’s $1 trillion AI gamble”

6–8 minutes read

News from Science
Made to order bioweapon? AI-designed toxins slip through safety checks used by companies selling genes
Researchers expose flaws in safeguards intended to block rogue actors from making bioweapons. Microsoft bioengineer Bruce Wittmann….used AI tools like a would-be bioterrorist: creating digital blueprints for proteins that could mimic deadly poisons and toxins… Wittmann and his Microsoft colleagues wanted to know what would happen if they ordered the DNA sequences that code for these proteins from companies that synthesize nucleic acids….a “red team” exercise, looking for weaknesses in biosecurity practice….according to their new paper, published today in Science, one key guardrail failed.
October 2, 2025
_____________________________

News from Observer
The $300 Billion A.I. Infrastructure Crisis Hiding in Plain Sight
As A.I. accelerates, a neglected $300 billion infrastructure gap threatens to slow its progress. The race to scale artificial intelligence has triggered historic investment in GPU infrastructure. ….This may be the largest corporate resource reallocation in modern history.
October 3, 2025
_____________________________

News from Bloomberg
Assume an AI Bubble. What Difference Would It Make?
Investors suffered, but the internet didn’t go away after the dot-com bust…. September saw the publication of at least two widely read analyses suggesting that what is going on in artificial intelligence is indeed a bubble ….A quarter-century has now passed since dot-com bombed; does this mean that its lessons have not been learned?
September 30, 2025
_____________________________

News from TechCrunch
Meta plans to sell targeted ads based on data in your AI chats
…. Now Meta will also use data from conversations with its AI chatbot … giving it another powerful signal to target its ads.
October 1, 2025
_____________________________

News from MIT Technology Review
OpenAI is huge in India. Its models are steeped in caste bias.
When Dhiraj Singha began applying for postdoctoral sociology fellowships in Bengaluru, India, in March, he wanted to make sure the English in his application was pitch-perfect. So he turned to ChatGPT. He was surprised to see that in addition to smoothing out his language, it changed his identity—swapping out his surname for “Sharma,” which is associated with privileged high-caste Indians.
October 1, 2025
_____________________________

News from Rest of World
AI is reshaping childhood in China
The one-foot tall AlphaDog, which weighs about eight kilos (18 pounds) was developed by robotics startup Weilan and is powered by DeepSeek’s AI model. In addition to practicing English with Wu’s son, it chats with him about current events, dances to his guitar music, and, through its built-in camera, helps Wu monitor the home when she is away.
October 1, 2025
_____________________________

News from The Hollywood Reporter
Does Sora 2 Come to Hollywood in Peace — Or to Leave It in Pieces?
…. A tool that will let users prompt … videos into existence and share on their social feeds. Now instead of dunking on people with off the rack memes…. The AI-ification of our TikTok feed is upon us.
September 30, 2025
_____________________________

News from Observer
A.I. Is Changing What Venture Capitalists Invest In and How They Invest
Investors say A.I. is not just what they’re funding but how they’re funding it.
With A.I. leaders like Sam Altman warning of a potential bubble, it might seem logical for investors to pull back. Instead, venture capitalists say they’re doubling down
September 30, 2025
_____________________________

News from NIST
CAISI Evaluation of DeepSeek AI Models Finds Shortcomings and Risks
AI models from developer DeepSeek were found to lag behind U.S. models in performance, cost, security and adoption. Security shortcomings and censorship may pose risks to application developers, consumers and U.S. national security.
September 30, 2025

_____________________________

News from Bloomberg (unlocked link)
OpenAI Releases Social App for Sharing AI Videos From Sora
OpenAI is releasing a standalone social app for making and sharing AI-generated videos with friends, an attempt to supercharge adoption for the emerging technology just as ChatGPT did for chatbots three years ago. The free Sora app, available Tuesday by invitation, is powered by a new version of OpenAI’s video-making software of the same name.
September 30, 2025
_____________________________

News from Wonkette
Creepy Right Wing Men Swoon Over Tilly Norwood, AI Actress ‘Without Opinions’
SAG-AFTRA, actors and normal human beings are less enthused….a lot of drama over “AI actress” (ie: “a cartoon”) Tilly Norwood making its debut at a film festival in Zurich…. is not an actor, it’s a character generated by a computer program that was trained on the work of countless professional performers — without permission or compensation.
October 2, 2025
_____________________________

News from Gizmmodo
OpenAI Rolls Out ChatGPT’s Ability to Buy Stuff for You
OpenAI’s new Instant Checkout lets users buy products from Etsy and soon millions of Shopify merchants….without leaving the chat….This move puts OpenAI one step closer to its bigger goal of creating a fully functional AI agent.
September 29, 2025
_____________________________

News from Semafor
AI could spur millions of layoffs, Kelly warns
“If we don’t focus on workers and think about this, about what does it mean to work in our country, and don’t put some guardrails in place and start some programs to train people for new jobs … we could see millions of people laid off here by the end of the decade,” he [Sen. Mark Kelly] said
September 30, 2025
_____________________________

News from Oligarch Watch
Sam Altman’s $1 trillion AI gamble
For at least a year, it’s been obvious there isn’t enough revenue in AI. When you add it all up—software, AI compute, GPUs for training and running models—it’s about $55 billion. ….half a trillion dollars invested ….OpenAI loses billions, Anthropic loses billions, Microsoft loses billions….circular financing—OpenAI spending heavily on Nvidia products, while Nvidia invests back into OpenAI….Nvidia essentially buys its own GPUs and then rents them back to OpenAI.
September 30, 2025

More news briefs —> 

#                        #                      #

Author