The A.I. Daily
Failed AI IPO Puts Fear Before FOMO
Breakingviews argues that a failed AI company IPO has put investor fear ahead of FOMO, potentially affecting the pace of future AI listings.
Industry
Everything · newest firstModels, products and AI startups, near-duplicates collapsed to the most credible source
OpenAI APAC Policy Head to Leave After Six Months
OpenAI's most senior Asia-Pacific public policy executive, Sanghyun Lee, is leaving after only six months. Based in Singapore, he joined OpenAI in April from Google. The departure comes as OpenAI faces scrutiny from regulators in the region.
Berkeley Lab boosts AI for particle accelerators
Berkeley Lab reports another boost in applying artificial intelligence to particle accelerators. The work uses AI to improve accelerator operation and control, showing AI's growing practical value in scientific facilities.
Principal's Shah on AI capex cycle and markets
Seema Shah, Chief Global Strategist at Principal Global Investors, discussed the AI capex cycle and its impact on investor sentiment with Bloomberg's Avril Hong at the Milken Institute Asia Summit in Singapore.
Opinion: 3 dilemmas on keeping AI under control are converging
A South China Morning Post opinion piece argues that three dilemmas around keeping AI under control are converging, spanning tensions among regulation, technological progress and geopolitical competition. It says these conflicts are increasingly intertwined, complicating AI governance.
LemonSeed: Qwen3.8-27B Hits 159 tok/s on iPad via eGPU R9700
LemonSeed Studio is an iPad editor that runs on-device inference via an AMD GPU in a Thunderbolt enclosure, embedding the upstream Linux amdgpu driver. Running Qwen3.8-27B Q4 with a DFlash2 draft, the R9700 decodes at 158.5 tok/s on iPadOS and 159.4 tok/s on macOS, versus a 31.2 tok/s baseline, while Strix Halo reaches 64.4 tok/s on Linux.
ThinkingBox benchmark: one-shot agent success rarely repeats
Microsoft authors released ThinkingBox-Bench: 507 business workflows across 5 domains, each run 20 times and graded on terminal backend state. Kimi-K3 solved 93.89% of tasks at least once but only 13.41% all 20 times, while Claude Opus 5 discovered fewer (79.09%) yet repeated far more (47.53%), showing discovery and repeatability rank models almost oppositely.
Will OpenAI Revenue Reset Stop AI Demand?
Bloomberg Intelligence's Mandeep Singh says that despite questions over OpenAI's revenue accounting, booming token growth points to solid AI infrastructure demand.
Qwen 3.6 35B praised for general agentic tasks
A user praised Qwen 3.6 35B (Unsloth UD_Q4_K_XL quant) for consistent general agentic tasks like smart-home control and messaging, reaching ~120-140 t/s generation on Tesla P100s, though it is weak at coding and holding a persona.
Nvidia-Backed Lumentum Sees Opto-Parts Sold Out to 2029
Nvidia-backed Lumentum says its optoelectronic parts are almost completely sold out through 2029 due to strong demand from AI data centers.
Stanford PhD student uses AI to design virus genetic blueprints
Stanford PhD student Samuel King used a generative AI model to propose genetic blueprints for microscopic viruses, not yet AI-generated life but possibly the next step. MIT Technology Review will livestream an interview with King by AI reporter James O'Donnell on Oct 16, discussing his work and his selection as an Innovators Under 35 honoree.
Edi Life OS: Self-Hosted Dashboard with MCP Server
Edi Life OS is a self-hosted life dashboard that unifies ten workspaces—focus, habits, goals, money and projects—into one private dashboard, with a built-in MCP server so AI assistants can work alongside users. It runs on any PHP+MySQL host with no subscription or tracking.
Australian music industry warns on AI copyright weakening
Australia's music industry warned at a parliamentary hearing that weakening copyright protections to attract OpenAI and Anthropic could turn the country into a 'Trojan horse for Big AI'. ARIA and PPCA chief executive urged AI developers to negotiate licensing deals rather than seek changes to copyright law.
Mathematician criticizes OpenAI's Partition Principle announcement
A mathematician harshly criticized OpenAI's preprint claiming the Partition Principle does not imply the Axiom of Choice, calling it poor and explaining why Sam Altman should not receive a promised bottle of whisky. The author also reflects on the need for a framework for AI use in mathematics.
Trump funding threat stalls state AI regulation bills
Several Louisiana AI regulation bills were shelved after a Trump executive order threatened to pull broadband funding from states that 'excessively' regulate AI, including bills requiring human review of AI-denied healthcare claims and banning AI robocalls. New federal broadband funding raises questions about the state's ability to regulate AI on its own.
Strata accused of rewriting GitHub history to hide Claude authorship
A Reddit user found that every historical commit in the Strata project was rewritten to strip 'Co-Authored by Claude' text, causing the built-in UPDATE script to fail due to no common ancestor. The user believes this was intended to conceal Claude's role in authoring the project.
Trump unveils $2.4B in tech giant AI pledges
President Trump announced $2.4 billion in computing commitments from tech giants including Nvidia, OpenAI and Anthropic for federal AI-driven scientific research, with Nvidia contributing $1 billion. The funding will expand the Genesis Mission to 14 federal agencies in energy, health and space, as Trump pushes to rebrand AI as 'Super Intelligence.'
Anthropic launches free AI security scans for open-source projects
Anthropic launched OSS Scanner, offering opt-in open-source projects free periodic security scans by its strongest models. Reports are fully model-generated with no human review, so they may be inaccurate, but scans can flag vulnerabilities faster.
New Hampshire proposes rules for AI in pharmacy
The New Hampshire Board of Pharmacy proposed rules under Part Ph 522 setting standards for pharmacy use of AI in prescription processing, drug utilization review, dispensing and clinical decision support. The core principle: technology may assist practice but cannot replace licensed personnel's professional judgment and accountability. It could be among the first state pharmacy-specific AI frameworks.
Washington leaders call for new AI regulations
Washington Sens. Patty Murray and Maria Cantwell are pushing AI rules at state and federal levels covering consumer protection, industry oversight and security risks. The state AI task force urged firms to disclose risk-mitigation plans, and a chatbot protection law takes effect in 2027.
OpenAI's math advances spark reckoning for academia
Bloomberg reports that OpenAI's advances in math are forcing some academics to rethink the future of mathematics, with one mathematician reflecting on how AI is changing his profession.
MIT Scholars Discuss AI's Path Forward in Academia
Sasha Rakhlin, director of MIT's Statistics and Data Science Center, published an essay on how AI is changing academia, noting that verification speed and reliability determine AI progress in specific disciplines.
Inside Elon Musk's midterm spending spree
WIRED's podcast discusses Elon Musk's America PAC spending millions on the midterms, an ideology Trump calls dangerous, and backlash against the designer of Meta's new AI logo.
Developers build simulations with frontier AI agents and Omniverse
NVIDIA highlights how developers combine frontier AI models such as GPT-6 Astra with Omniverse libraries to build simulation applications via natural-language instructions. Examples include an interactive warehouse humanoid simulator and an autonomous-driving testing workflow, with Omniverse supplying GPU-accelerated physics, rendering and sensor simulation.
Bloomberg documentary explores how AI is sucking up global wealth
A Bloomberg Originals documentary argues that the AI boom has concentrated immense wealth and pushed global trade imbalances to levels unseen since the 2008 financial crisis.
Rembrandt, an open source local AI photo editor, launches
The open source project Rembrandt launched as a free, fully local alternative to Adobe Lightroom for macOS, Windows and Linux, with AI features such as natural-language edits, AI subject and depth masks and 2x/4x GPU super resolution. It needs no account or subscription and can be self-hosted for browser-based editing.
Engineer builds Theranos desk simulator from trial documents
Extend engineer Bo Lau used over a thousand emails, slides and documents released in the US v. Elizabeth Holmes trial to build a website simulating the Theranos CEO's 2016 desk, letting users browse court exhibits on a virtual iPhone and MacBook and even run the Edison blood-testing device. The project is partly a joke and partly a way to navigate case files, while promoting a Holmes documentary watch party.
DOE, NIH and Biohub sign pact on biology superintelligence
The US Department of Energy, the National Institutes of Health and Biohub signed an agreement to collaborate on 'biology superintelligence' research. Details of the pact were not disclosed.
Have URMs and UTs been adopted by frontier models?
A Reddit discussion questions whether frontier labs at big tech have adopted Universal Reasoning Model (URM) and Universal Transformer (UT) enhancements, as UT-based small models trained from scratch outperform standard LLMs on certain tasks by a significant margin.
California commission moves to shut down human vs. robot cage fights
The California State Athletic Commission sent a cease-and-desist letter to startup Rek, which hosted a September 18th match pitting human Frankie LaPenna against a humanoid robot piloted remotely via a VR system. Rek calls itself a humanoid robot fighting league, and the robot appears to be an EngineAI unit fitted with a Terminator-like head.