artificial intelligenceFriday, October 9
Two stories cut through the noise today. OpenAI caught state actors using ChatGPT for influence operations, and a major AI datacentre IPO collapsed. Together they paint a picture of an industry grappling with both geopolitical weaponisation and investor skepticism.
The day's most consequential story is about AI being used in the real world, not just in theory.
OpenAI banned Russian and Iranian accounts that used ChatGPT to generate false narratives that got picked up by news outlets and even prompted responses from European Parliament members. This is one of the clearest examples yet of generative AI being used in state-backed influence campaigns.
OpenAI's text watermark detection drops to 17% after 25% of words are changed, according to internal tests. The company is rolling out watermarks in the EU to comply with the AI Act but restricting the detector to approved researchers, which raises questions about how useful the tool really is.
Two physicists claim to have found a mathematical formula that predicts when an AI system is about to generate harmful outputs. If it holds up, this could give developers a way to catch dangerous responses before they happen, especially on offline systems.
Anthropic updated its usage policy to ban sustained abusive or cruel behavior toward Claude. It's a minor policy tweak, but it signals that even the safety-conscious lab is thinking about how users treat models.
Meanwhile, the economics of building all that infrastructure are getting a second look.
Firmus, the Nvidia-backed Australian datacentre company, pulled its IPO citing insufficient investor demand. It was set to be the biggest ASX debut since Telstra, and the failure suggests the market is cooling on the massive capital spending required by AI infrastructure.
Paul Krugman argues that the immediate economic impact of AI isn't job displacement or a bubble, it's hyperscalers crowding out other investment with their datacentre spending. A useful lens for understanding the Firmus IPO failure.
India's push to become an AI data centre hub is facing local pushback, with residents protesting construction that disrupts their neighborhoods. The tension between national ambition and community impact is playing out globally.
“The AI Doc: Or How I Became an Apocaloptimist”: Daniel Roher, Tristan Harris Discuss New Filmwww.democracynow.org
What Is HAIA-SMART? The Social Media AI Rating Toolbasilpuglisi.com
GitHub - edrisranjbar/lifeos: Self-hosted life OS: habits, goals, Kanban, finance and focus, with an MCP server for AI.github.com
André Staltz — Time Till Open Source Alternative is now zerostaltz.com
CoreWeave targets AI inference bottlenecks with full-stack optimizationsiliconangle.com
Grimes Shares First Song From AI-Inspired Album Psy Operapitchfork.comPresident Trump announced $2.4 billion in computing power commitments from major tech companies, including Nvidia ($1 billion), AMD, OpenAI, and Anthropic, to support the "Genesis Mission." This initiative aims to deploy AI-driven scientific research across 14 federal agencies in
Fan Bingbing Questions AI Microdramas, Praises Park Chan-wook at Busanvariety.com
Chinese actress Fan Bingbing attended the Busan International Film Festival to present films by Wayne Wang and Ivan Tan. During a press conference, she expressed skepticism about the value of AI-generated microdramas for young audiences in China. She also stated her desire to col
Roundtables: A Conversation With the Creator of AI-Designed Viruseswww.technologyreview.com
MIT Technology Review invites subscribers to a live roundtable interview with Samuel King, a Stanford PhD student and Innovator Under 35, who used generative AI to propose genetic blueprints for microscopic viruses. The discussion covers his work on AI-designed biology and its im
AI-Produced Series ‘Space Nation’ and ‘Kira’ to Premiere at Mipcomvariety.com
Two science-fiction series, "Space Nation" and "Kira," produced entirely with generative AI by Ex Machina Studios, will premiere their pilot episodes at the Mipcom market in Cannes. K5 Intl is handling worldwide sales. The companies claim this marks the first time fully produced
GEMS plans AI-led learning in schools from 2027www.tes.com
GEMS Education announced the "GR8 School" initiative, an AI-led learning model set to launch in 2027. The program is built on three pillars, starting with "hyper-personalised learning," which allows students to progress through core academic material at their own pace using AI. A
Nvidia-backed Aussie AI firm Firmus withdraws historic IPO, citing market volatilitywww.cnbc.com
Firmus, an Australian AI data center operator backed by Nvidia, has withdrawn its planned initial public offering (IPO) due to market volatility and unfavorable terms. The company's board determined that proceeding with the IPO was not in the best interest of shareholders and sta
Trump says anyone who does not refer to AI as ‘super intelligence’ is ‘THE ENEMY' — POLITICOapple.news
Donald Trump claims that anyone who refuses to refer to AI as "super intelligence" is "THE ENEMY," following an executive order he signed a week prior to rename the technology.
finance roundup
education roundup
health roundup
personal roundup
books roundup
music roundup
