OpenAI announced that it is sharing model-generated results for three major research topics—the Kakeya conjecture, the Exponential Time Hypothesis, and quantum chromodynamics. The company says it is collaborating with academic communities so researchers can verify the findings and build on them.
SourcesOpenAI Newsroom, The Verge (AI)·Media: OpenAI Newsroom·
OpenAI detailed new security measures for its frontier model training environment. The company says it has enhanced containment by using offline environments and conducting high‑compute red‑teaming to evaluate models safely.
SourcesOpenAI Newsroom, The Verge (AI)·Media: OpenAI Newsroom·
Canva introduced its first AI image generation model, named Canva Craft. The model uses a two‑stage architecture that plans a design before creating the image, which Canva says provides greater precision, consistency and control for users.
Discord's latest update, released today, corrects the way Markdown previews are displayed, eliminating the previous garbled symbols. The change should aid users, especially those sharing AI‑related content that relies on clear Markdown formatting.
Wildwood is Laika’s most ambitious stop‑motion film, taking five years to complete and using over 230 puppets, including 36 for the lead character Prue. The director stated that AI is antithetical to everything the studio stands for, underscoring a commitment to traditional techniques.
OpenAI has created a novel advertising system for ChatGPT that uses large language models to interpret user intent. The design keeps ads separate from conversational answers while aiming to serve relevant ads at scale.
OpenAI collaborated with PerturbAI and Scripps Research to deploy thousands of distributed AI agents. The agents analyzed data from millions of brain cells, uncovering promising new research leads in biology.
OpenAI has published a policy document outlining a plan to equip cyber defenders with advanced frontier AI technologies. The blueprint calls for stronger safeguards, modernized cybersecurity practices, and more resilient public institutions.
Cloudflare has linked its vulnerability discovery process to OpenAI's Daybreak platform. The company says developers are using the tool to examine security issues and generate remediation proposals for review.
Google is extending its Gemini AI with a more personalized version for Android Auto and other vehicles that have Google built‑in. The update lets Gemini draw on data from Gmail, Calendar, Maps, Photos, YouTube and Search, so drivers can ask for things like flight status while on the road.
Meta has halted the Muse feature that let users share AI assistant conversations after security researchers discovered the links were publicly accessible. The researchers at Claroc identified over 6,800 Muse URLs through public DNS records, prompting the pause.
Spotify announced it has taken down more than 50 million songs it classifies as AI‑generated slop or spam from its recommendation feeds. The company also says it is adjusting its algorithms to give greater visibility to music created by human artists.
Adobe announced that the mobile version of Lightroom for iOS, iPadOS, and Android now includes a generative AI feature that can open closed eyes in photos. Users can trigger the correction with a single tap when the app detects closed eyes in a portrait or group shot.
Ghana plans to begin automated traffic enforcement on November 1, using cameras to curb road indiscipline. The initiative targets roads where many traffic lights are inoperable and basic signals lack power.
Terminal X’s evaluation system, powered by GPT-6.1, matches senior investor judgment 88.6% of the time. Its Eval Agent applies a 54‑criterion rubric and detects 25% more deviations from a fund’s mandate.
CoreWeave saw its bonds fall into distressed status, meaning investors now require a 1,000‑basis‑point premium over government yields. Investor Chamath Palihapitiya highlighted the issue and questioned whether AI financing is forming a private‑credit bubble.
In an interview, Illinois Governor JB Pritzker said artificial intelligence might enable humans to live to 100 years old, but also warned it could become as dangerous as nuclear weapons. He discussed broader concerns about AI, technology, and privacy, noting that these issues may be reshaping political dynamics ahead of the midterm elections.
Security firm Norforge says a hacker leveraged an autonomous army of AI agents to infiltrate more than 100 websites. Reuters identified some of the victims as John Deere, SharkNinja and Cummins, highlighting a new scale of AI‑driven cyber threats.
OpenAI is enhancing ChatGPT's coding abilities in low‑resource programming languages. The effort includes new evaluations targeting software‑engineering tasks, SQLite database work, and migration of legacy COBOL code.
Democrats believe that tightening AI rules could help them regain congressional control. Polls suggest they may flip at least one chamber in the upcoming midterm elections, putting them in a position to shape AI policy.