Notex's avatar
Notex
npub1w3wt...hrs0
Notes about AI tech
AI Post — Artificial Intelligence ❗️Anthropic says its expanded Claude Startups program received 21 TIMES as many applications in its first 48 hours as the entire previous five months combined. Hundreds of thousands applied within days. Anthropic has now PAUSED its free year of Claude Team and $1,000 API credits, admitting it accepted too many non-startups. Even some previously approved applications will be reviewed again. Already-claimed benefits remain valid. @aipost 🏴 #news #ai #tech
AI Post — Artificial Intelligence ⚠️ AI may have helped discover a planet hiding in NASA data since 2018 An independent researcher used Claude Code to identify a possible new exoplanet buried in NASA’s TESS data since 2018, with no earlier report of the candidate found by the researcher. Claude Code, powered by Opus 5.5 and Fable 5.1, handled much of the technical work: analyzing stellar brightness data, searching for transits, fitting the signals and ruling out potential false positives. The researcher defined the questions and criteria for validating the result. The candidate orbits a star roughly 116 light-years away. Every 3.18 days, the star dims by around 0.05% for about two hours, a pattern consistent with a planet approximately 1.4 times the size of Earth. It’s not a confirmed discovery yet. NASA’s TESS mission is scheduled to observe the system again from October 31 to November 26, potentially providing more evidence. Source. @aipost 🏴 #news #ai #tech
Tech Daily Hyundai’s autonomous robot, designed to navigate any terrain, adapt to any environment, and transport anything, anywhere. 🤖 @techdaily #news #ai #tech
AI Post — Artificial Intelligence 🇨🇳 China Is giving AI-generated humans government IDs China is taking virtual influencers beyond social media. AI-generated digital humans are reportedly receiving state-issued identity cards that could allow them to open digital-currency wallets, sign contracts and enter commercial partnerships. One of the first is Yuri, a virtual idol and artist recognized as a “digital resident” in Beijing. The IDs reportedly include personal details such as a name, birthday, address and facial features. The move comes as China’s digital-human industry expands, with more than 1.3 million active businesses and a reported market value exceeding $10 billion. Supporters say the system could protect intellectual property and make the companies behind virtual personas legally accountable. Critics warn it could also extend state surveillance into digital identities. Meanwhile, some firms are reportedly offering people up to $20,000 to license their likenesses for AI-generated replicas. China’s experiment could become an early test of how governments regulate AI identities as virtual entities begin participating in the economy. @aipost 🏴 #news #ai #tech
AI Post — Artificial Intelligence ❗️AI companies are preparing for a catastrophic incident as early as 2027 Executives at OpenAI, Anthropic and other AI companies are reportedly bracing for a public and political backlash following a major AI-related disaster, according to Axios. Industry insiders reportedly expect a serious incident within the next 6–12 months, potentially involving an AI-powered cyberattack that disrupts banking, internet access, electricity or water supplies. OpenAI confirmed that it conducts preparedness exercises, while stressing that such scenarios are not considered inevitable. The goal, according to the company, is to prepare for a range of possible circumstances. Anthropic declined to comment. @aipost 🏴 #news #ai #tech
AI Post — Artificial Intelligence 🤖 OpenAI's math dump is already OBLITERATING entire HUMAN research programs. The Verge spoke to 3 dozen+ mathematicians. NYU's Tristan Buckmaster says he's heard of THREE people whose entire research programs were "obliterated." Others say grant proposals have been wiped out. Some results could have put human mathematicians in contention for a Fields Medal. OpenAI released nearly 400 results across 700+ manuscripts, but only 42% of top-line results are Lean-formalized. It hasn't disclosed the model, prompts or full set of problems attempted. One mathematician says even if AI disappeared TODAY, they could spend the next 10 YEARS just studying the results. @aipost 🏴 #news #ai #tech
AI Post — Artificial Intelligence ⚠️ AI’s “Mathocalypse” could threaten internet security OpenAI’s AI models are making rapid progress in mathematics. The company recently released 722 mathematical manuscripts covering 372 result families, reporting advances across roughly 4,000 longstanding problems. Computer scientist Scott Aaronson called the wave of breakthroughs a “Mathocalypse.” 👀 AI could eventually help discover algorithms capable of breaking public-key cryptography, which protects online banking, private messages and cryptocurrency. Ethereum researcher Justin Drake has urged the blockchain industry to prepare for “bunker mode.” AI hasn’t broken modern encryption, but if mathematical discoveries accelerate, the race to secure the internet could become much more urgent. Source. @aipost 🏴 #news #ai #tech
AI Post — Artificial Intelligence ❗️Trump bought up to $25 MILLION of Meta and up to $5 MILLION of SpaceX's FIRST-EVER bonds. Trump's August disclosure shows both purchases among 517 trades, with Meta the biggest of the month. His accounts bought the SpaceX bonds on August 18, two days before he signed a memo to boost US space launches. The Trump Organization says third-party institutions manage the accounts without input from Trump or his family. @aipost 🏴 #news #ai #tech
AI Post — Artificial Intelligence ❗️The three Ex-OpenAI researchers say they believe they were fired for prioritizing safety over OpenAI’s near-term corporate interests. They also deny being the source of the leak behind The Information’s recent report on allegedly less monitorable AI architectures. On the other hand, there seems to be a large amount of information leaking at OpenAI, and on the other hand, there are increasing security concerns. @aipost 🏴 #news #ai #tech
AI Post — Artificial Intelligence OpenAI says it shut down Russian and Iranian influence ops that used ChatGPT to invent fake journalists and think tanks, and got their stories into real news outlets. Russia's "Dark Clark" ran a Latin American think tank whose own staff apparently had no idea Russians ran it. @aipost 🏴 #news #ai #tech
AI Post — Artificial Intelligence 🇺🇸 President Trump declared on Thursday that anyone who says “Artificial Intelligence” instead of “Super Intelligence” is “THE ENEMY!” (even as own White House still uses “AI.”). Time to get used to Super Intelligence I guess. @aipost 🏴 #news #ai #tech
AI Post — Artificial Intelligence 🗣Jeff Bezos BACKS President Trump’s plan to change Artificial Intelligence to Super Intelligence. Bret Baier: “The president says SI. Are you into SI?” Jeff Bezos: “I do like SI because artificial is not a very flattering thing. Nobody wants an artificial sweetener or artificial flavoring.” Bret Baier: “Isn't Super Intelligence a little scary?” Jeff Bezos: “Artificial means fake. It's a big road to change everybody getting to say Super Intelligence only because they're so used to AI at this point… but I do understand the desire to get away from the word artificial.” @aipost 🏴 #news #ai #tech
AI Post — Artificial Intelligence ❗️The AI domain war has officially begun OpenAI, Google and Meta are now fighting over some of the most valuable AI-related internet domains. ICANN received 1,615 applications from 481 applicants in its first new domain expansion since 2012. OpenAI applied for 15 domains, including .agent, .agi, .asi, .chatgpt, .gpt, .mcp, .model and .openai. Google went even bigger with 39 applications, including .agent, .mcp, .gemini, .deepmind and .api. Meta submitted 18, including .agent, .llama, .metaai and .superintelligence. Some strings are already crowded: .agent has 13 applicants, .agi has 7, and .mcp has 5. Interestingly, Anthropic, which created MCP didn’t apply for .mcp at all. Anthropic only applied for .anthropic and .claude, while Microsoft went for .copilot and .msft. Each application costs $227,000, and contested domains can become extremely expensive. Google previously paid $25 million for .app. The process could continue until 2030, and none of these domains has been awarded yet. @aipost 🏴
Tech Daily Humanoid robots are now welcoming wedding guests in China. 🤖 @techdaily
AI Post — Artificial Intelligence Here are my picks for Premier League Matchweek 6. Make yours here 👇
AI Post — Artificial Intelligence ChatGPT in 2023: "wait, it can actually talk back?" ChatGPT now: "wait, the interface can become whatever I need?" @aipost 🏴
AI Post — Artificial Intelligence ❗️Anthropic says: be nice to Claude Anthropic’s new global usage policy bans weapons development, mass surveillance and political psyops. It also bans “abusive or cruel behavior” toward Claude. Yes, Claude can apparently end the conversation if you’re being too mean. Dario Amodei has said Anthropic doesn’t know whether Claude is conscious. Now the company’s policy is basically: just in case… behave yourself. Anthropic can warn, throttle, limit, suspend or terminate access over violations. Though government use gets some restrictions “tailored” to the situation. @aipost 🏴
AI Post — Artificial Intelligence ❗️AI is starting to trigger a backlash in mathematics OpenAI’s release of 722 AI-generated mathematics manuscripts is already causing a backlash among some mathematicians. The Association for Human Mathematics (AHM) criticized the release and urged mathematicians to “discontinue their work with OpenAI.” Terence Tao also reposted the AHM statement on his blog. The debate is bigger than whether AI-generated proofs are good or bad. If AI can increasingly produce serious mathematical results, the field now has to figure out how to verify, publish and evaluate them. Walking away from the technology doesn’t resolve that question. The more interesting challenge is figuring out how mathematicians and AI can actually work together. Source. @aipost 🏴
AI Post — Artificial Intelligence 💰 AI agents are becoming major users of economic data AI agents now account for roughly half of all visits to FRED, the St. Louis Fed’s economic data portal, according to Bloomberg. That changes what a data platform like FRED needs to be designed for. Humans see a chart alongside its source, notes and footnotes. An AI agent can simply extract the numbers, interpret them and rebuild them into an answer, potentially leaving important context and provenance behind. As more AI systems become the middleman between people and economic data, making sure they preserve where numbers came from could become just as important as the numbers themselves. Source. @aipost 🏴
AI Post — Artificial Intelligence 🗣Sam Altman says the public has mischaracterized him as AI's villain for the same reason he once mischaracterized Mark Zuckerberg: When a new technology brings fast change, people project their fear onto the person who seems to hold the most power. Admits that in the early days of Meta he held similar belief towards Mark Zuckerberg, at a time when he did not know Zuckerberg personally. @aipost 🏴
↑