AI News Flash

Епизоде

• 
  [10/18 21:00] Anthropic commits 5% of its shares to a new public benefit fund that grows with the company's value / New Zealand Police fed 350,000 lines of seized Chinese audio into Google Translate, documents show 18.10.2026 4мин

  AI News Flash for October 18, 2026 at 21:00 UTC. Topics: 1. Anthropic commits 5% of its shares to a new public benefit fund that grows with the company's value https://www.anthropic.com/news/public-benefit-fund 2. New Zealand Police fed 350,000 lines of seized Chinese audio into Google Translate, documents show https://www.rnz.co.nz/news/crime-and-justice/2010561/police-ai-investigator-uploads-350-000-lines-of-audio-in-risky-workaround 3. Australia's spy agencies face an AI legal vacuum, the national security laws monitor warns https://www.canberratimes.com.au/story/9368764/australian-spy-agencies-face-scrutiny-over-unregulated-ai-use/ 4. Israeli children's hospital launches an AI tool that helps nonverbal children communicate https://www.i24news.tv/en/news/israel/society/artc-israeli-hospital-launches-ai-tool-to-help-nonverbal-children-communicate 5. AI is closing the door on Bangladesh's female freelancers, The Daily Star reports https://www.thedailystar.net/slow-reads/unheard-voices/news/how-ai-closing-the-door-bangladeshs-female-freelancers-4301956
• 
  [10/18 18:00] Trump's intelligence chief says AI companies are responsible when their products hack websites / Iran's parliament speaker warns Gulf governments not to fund AI companies 18.10.2026 3мин

  AI News Flash for October 18, 2026 at 18:00 UTC. Topics: 1. Trump's intelligence chief says AI companies are responsible when their products hack websites https://www.foxnews.com/media/trump-intel-chief-warns-ai-firms-hacks-says-new-department-cop-beat 2. Iran's parliament speaker warns Gulf governments not to fund AI companies https://wanaen.com/ghalibaf-warns-regional-governments-against-supporting-ai-companies/ 3. Apple plans a higher-end HomeView with a 360-degree robotic arm, Bloomberg reports https://www.macrumors.com/2026/10/18/apple-rumored-to-launch-homeview-max/ 4. Chinese AI-made short dramas are booming in South Korea, especially among middle-aged men https://www.chosun.com/english/national-en/2026/10/19/IDFZFDJS6FCEXBHAS34GY4VRXA/ 5. Australian consumer group survey finds most people say AI made customer service worse https://aapnews.aap.com.au/news/ai-customer-service-giving-consumers-an-error-message
• 
  [10/18 15:00] Always-listening AI wearables are testing US state wiretap laws, and wearers may carry the risk / A Philippine House bill would set rules for transactions carried out by AI agents 18.10.2026 3мин

  AI News Flash for October 18, 2026 at 15:00 UTC. Topics: 1. Always-listening AI wearables are testing US state wiretap laws, and wearers may carry the risk https://www.cnbc.com/2026/10/18/ai-wearables-privacy-wiretap-laws.html 2. A Philippine House bill would set rules for transactions carried out by AI agents https://politiko.com.ph/villafuertes-push-guidelines-for-transactions-made-by-ai-agents/ 3. UK Prime Minister Andy Burnham seeks closer alignment with the EU on AI https://www.telegraph.co.uk/business/2026/10/18/burnham-seeks-closer-alignment-with-eu-on-ai/ 4. Saudi Arabia's Supreme Judicial Council adopts AI-based rules for preparing and reviewing cases https://saudigazette.com.sa/article/665525/saudi-arabia/supreme-judicial-council-adopts-rules-that-utilize-ai-technologies 5. Russian demand for programmers falls 47% and for testers 55% as language models take on routine code https://www1.ru/en/news/2026/10/18/spros-na-programmistov-i-testirovshhikov-v-rossii-rezko-upal-minus-47-i-55.html
• 
  [10/18 12:00] A second German court holds Google liable for its AI search answers / Google is negotiating to buy books from Korean publishers for AI training 18.10.2026 3мин

  AI News Flash for October 18, 2026 at 12:00 UTC. Topics: 1. A second German court holds Google liable for its AI search answers https://the-decoder.com/two-german-courts-now-hold-google-liable-for-what-its-ai-overviews-say/ 2. Google is negotiating to buy books from Korean publishers for AI training https://en.sedaily.com/international/2026/10/18/google-seeks-korean-book-deals-as-ai-copyright-bills-stall 3. Local data center moratoriums are barely slowing the US buildout, SemiAnalysis finds https://www.tomshardware.com/tech-industry/data-centers/300-plus-local-data-center-bans-have-only-delayed-just-three-us-projects-bans-much-more-bark-than-bite-only-2-3-gigawatts-of-data-centers-are-held-up-versus-38-gigawatts-expected-online-in-2027 4. US cybersecurity job postings fall 26%, while AI security engineer demand soars https://www.securityweek.com/us-cybersecurity-job-postings-fall-26-as-ai-security-engineer-demand-soars/ 5. Only 18% of Bay Area residents expect AI to benefit society, poll finds https://www.mercurynews.com/2026/10/18/survey-techies-more-optimistic-about-ai-than-other-bay-area-locals-but-still-worried/
• 
  [10/18 09:00] Drones hit a VK data center near St. Petersburg / Nvidia's custom HBM to use a TSMC-made base die, and Samsung goes along 18.10.2026 3мин

  AI News Flash for October 18, 2026 at 09:00 UTC. Topics: 1. Drones hit a VK data center near St. Petersburg https://www.kyivpost.com/post/87403 2. Nvidia's custom HBM to use a TSMC-made base die, and Samsung goes along https://en.sedaily.com/finance/2026/10/18/samsung-to-use-tsmc-for-base-die-in-nvidias-custom-hbm 3. Vance says data centers must be good community citizens https://www.cbsnews.com/news/jd-vance-cbs-town-hall-highlights-michigan/ 4. AI-generated nude images of schoolgirls in Malaysia https://www.nst.com.my/news/regional/2026/10/1558225/police-receive-five-reports-over-ai-generated-nude-images-300 5. IDA Pro maker Hex-Rays buys an Israeli AI security-research startup https://www.calcalistech.com/ctechnews/article/hytvgbf3ze
• 
  [10/18 06:00] Microsoft pulls out of Nscale's Essex data centre deal, a blow to the UK firm's planned listing / An AI-assisted dashboard helps detained protesters in New Delhi get home 18.10.2026 3мин

  AI News Flash for October 18, 2026 at 06:00 UTC. Topics: 1. Microsoft pulls out of Nscale's Essex data centre deal, a blow to the UK firm's planned listing https://www.telegraph.co.uk/business/2026/10/18/microsoft-axes-ai-deal-setback-uk-data-centre-pioneer/ 2. An AI-assisted dashboard helps detained protesters in New Delhi get home https://www.aljazeera.com/news/2026/10/18/ai-tool-helps-stranded-protesters-find-their-way-home-in-the-indian-capital 3. MIT model links 3D crystal structures with language to find new materials https://news.mit.edu/2026/giving-ai-crystal-clear-view-of-materials-world-1018 4. Japan's Neyagawa city launches a 24-hour generative AI phone line for older residents https://news.web.nhk/newsweb/na/nd-20261018de61097 5. South Korea tests a free national AI service ahead of a December launch https://www.afpbb.com/articles/-/3658115
• 
  [10/18 03:00] Rising financing costs hit Asia's AI supplier stocks; AI-linked perpetual futures slide over the weekend / Hyundai and Kia unveil a roadmap to autonomous, AI-run factories by 2032 18.10.2026 3мин

  AI News Flash for October 18, 2026 at 03:00 UTC. Topics: 1. Rising financing costs hit Asia's AI supplier stocks; AI-linked perpetual futures slide over the weekend https://www.investing.com/news/stock-market-news/asian-ai-stocks-fall-as-rising-financing-costs-threaten-spending-boom--bloomberg-4955727 2. Hyundai and Kia unveil a roadmap to autonomous, AI-run factories by 2032 https://biz.chosun.com/en/en-industry/2026/10/18/QLQSQOO5QNBBJIEYAPQTHFNCO4/ 3. AhnLab tests the AI hacking tool ARTEX, which broke into internal networks on its own https://www.asiae.co.kr/en/article/2026101809501237799 4. A Japanese city will let an AI character listen to schoolchildren's worries https://www.47news.jp/15098637.html 5. Jeffrey Epstein gave 100,000 dollars to fund AI research at the University of New Mexico in 2001 https://www.santafenewmexican.com/news/local_news/epstein-gave-100k-to-unm-for-artificial-intelligence-research-in-2001/article_53e7c2f9-9db1-4617-a3c6-b05cbaa90c17.html
• 
  [10/18 00:00] Publicly disclosed software flaws hit a record pace as AI speeds up discovery / Tech and AI companies pour money into U.S. state attorney general races 18.10.2026 3мин

  AI News Flash for October 18, 2026 at 00:00 UTC. Topics: 1. Publicly disclosed software flaws hit a record pace as AI speeds up discovery https://japantoday.com/category/tech/urgent-cybersecurity-flaw-discoveries-soar-to-record-high-with-use-of-ai-data 2. Tech and AI companies pour money into U.S. state attorney general races https://www.yahoo.com/news/politics/articles/big-tech-ai-firms-pour-213346961.html 3. AI coding agents write econometric code far more accurately, but silent errors remain https://cepr.org/voxeu/columns/letting-ai-agents-write-econometric-code 4. Intel, Renesas and Micron lay out chip plans for humanoid robots https://www.thelec.net/news/articleView.html?idxno=14667 5. Fear of AI copycats leads an indie game studio to cancel a public demo https://mashable.com/entertainment/drunken-goddess-reflux-steam-ai-demo-canceled
• 
  [10/17 21:00] OpenCode discloses breach that may have exposed users' stored AI provider keys / Sergey Brin says Google is getting on the RSI treadmill, while DeepMind's Argon safety report finds no significant acceleration yet 17.10.2026 3мин

  AI News Flash for October 17, 2026 at 21:00 UTC. Topics: 1. OpenCode discloses breach that may have exposed users' stored AI provider keys https://officechai.com/ai/opencode-discloses-security-breach-says-attackers-may-have-accessed-users-openai-anthropic-and-google-api-keys/ 2. Sergey Brin says Google is getting on the RSI treadmill, while DeepMind's Argon safety report finds no significant acceleration yet https://thenextweb.com/news/sergey-brin-gemini-recursive-self-improvement-safety-report 3. Meta removes the word "train" from its new scraping terms, but legal experts remain skeptical https://www.niemanlab.org/2026/10/legal-experts-warn-metas-new-terms-still-grant-broad-scraping-permissions-despite-language-changes/ 4. Anthropic president Daniela Amodei calls AI romantic partners a terrible use of AI https://thenextweb.com/news/daniela-amodei-ai-girlfriends-boyfriends-terrible-use 5. Latter-day Saint leader Elder Gong proposes "Humanity's First Exam" to test AI moral reasoning https://www.deseret.com/faith/2026/10/17/elder-gong-proposes-ai-moral-evaluation-humanitys-first-exam/
• 
  [10/17 18:00] SoftBank weighs $18 billion Allied Universal deal, eyes replacing some guards with robots / Texas promises a data center water crackdown, but the penalty tops out at $500 17.10.2026 3мин

  AI News Flash for October 17, 2026 at 18:00 UTC. Topics: 1. SoftBank weighs $18 billion Allied Universal deal, eyes replacing some guards with robots https://www.kslm.news/news/softbank-weighs-18-billion-acquisition-of-allied-universal-bloomberg-news-reports-mvbf93hc 2. Texas promises a data center water crackdown, but the penalty tops out at $500 https://www.yahoo.com/news/us/articles/texas-promises-data-center-water-172100238.html 3. Monday.com faces wave of senior exits amid AI-centered restructuring https://www.calcalistech.com/ctechnews/article/2wsa7rgfb 4. Malwarebytes warns phishing emails are baiting AI assistants with hidden instructions https://futurism.com/artificial-intelligence/ai-agent-sucker-phishing-emails 5. Japan turns to AI maps, robot wolves and drones to fight record bear attacks https://www.yahoo.com/news/world/articles/japan-deploys-robots-pepper-spraying-133936830.html
• 
  [10/17 15:00] Websites are closing their doors to AI agents: Yelp, Delta, JPMorgan, Schwab / Brazil's electoral court orders Meta to treat Lula and Bolsonaro equally, including in Meta AI 17.10.2026 3мин

  AI News Flash for October 17, 2026 at 15:00 UTC. Topics: 1. Websites are closing their doors to AI agents: Yelp, Delta, JPMorgan, Schwab https://www.theverge.com/ai-artificial-intelligence/1013195/ai-agents-blocked-web 2. Brazil's electoral court orders Meta to treat Lula and Bolsonaro equally, including in Meta AI https://www.wral.com/news/ap/42270-brazils-lula-challenges-meta-at-electoral-court-for-alleged-algorithmic-bias-and-ai-restrictions/ 3. Gruffalo illustrator Axel Scheffler considers legal action over AI-generated images https://www.yahoo.com/news/world/articles/gruffalo-illustrator-intends-legal-action-141826067.html 4. Taiwan's defense ministry buys an AI satellite imagery system to track PLA forces https://www.scmp.com/news/china/military/article/3371252/how-taiwans-new-ai-satellite-system-aims-speed-early-warning-against-pla-moves 5. AI midwife chatbot DawaMom fills gaps in Zimbabwe's maternal health care https://www.rfi.fr/en/africa/20261017-ai-midwife-seeks-to-fill-gaps-in-zimbabwe-s-maternal-health-care
• 
  [10/17 12:00] IMF statement calls AI "advanced intelligence technologies" after US push for "super intelligence"; Anthropic's Jack Clark speaks in Bangkok / DeepSeek releases V4.1-Flash-1017, scoring 77.0 on DeepSWE in its own table, above Claude Opus 5 17.10.2026 3мин

  AI News Flash for October 17, 2026 at 12:00 UTC. Topics: 1. IMF statement calls AI "advanced intelligence technologies" after US push for "super intelligence"; Anthropic's Jack Clark speaks in Bangkok https://www.cnbc.com/2026/10/17/imf-members-call-for-more-support-amid-war-shocks-advanced-intelligence-risks.html 2. DeepSeek releases V4.1-Flash-1017, scoring 77.0 on DeepSWE in its own table, above Claude Opus 5 https://www.marktechpost.com/2026/10/17/deepseek-just-released-v4-1-flash-1017-scores-77-0-on-deepswe-opens-search-api/ 3. Senior UK Labour figures name AI's power as a top worry; former AISI chief scientist warns open models could already do serious damage https://www.theguardian.com/technology/2026/oct/17/artificial-intelligence-ai-labour-government-safety-uk-extremism 4. Mathematicians dispute OpenAI's account of its Navier-Stokes proof; the role of its 10,000-agent swarm is unclear https://the-decoder.com/mathematicians-say-openai-doesnt-understand-its-own-navier-stokes-proof/ 5. Croatia unveils a national AI plan to 2032 worth 932 million euros https://www.croatiaweek.com/croatia-artificial-intelligence-plan-2032/
• 
  [10/17 09:00] Anthropic studies its own workforce; junior share of technical hires fell from 27% to 9% / South Korea's deepfake sex-crime reviews top 17,000 in eight months, beating last year's total 17.10.2026 3мин

  AI News Flash for October 17, 2026 at 09:00 UTC. Topics: 1. Anthropic studies its own workforce; junior share of technical hires fell from 27% to 9% https://www.anthropic.com/research/employment-at-anthropic 2. South Korea's deepfake sex-crime reviews top 17,000 in eight months, beating last year's total https://en.sedaily.com/society/2026/10/17/deepfake-sex-crime-reviews-top-17000-in-eight-months 3. Google launches Universal Cart across Search and Gemini in the US; big retailers offer only select checkout https://startupfortune.com/googles-universal-cart-goes-live-and-retailers-are-already-pushing-back/ 4. China Meteorological Administration puts Fengqing AI ensemble forecast system into operation https://pandaily.com/cma-fengqing-ai-global-ensemble-weather-forecast-system-v1-0-operational-11-day-skill 5. Kuaishou launches Kling 4.0 video model with 30-second single-pass generation https://macaubusiness.com/kling-ai-officially-launches-next-generation-kling-4-0-raising-the-bar-for-professional-grade-video-models/
• 
  [10/17 06:00] Anthropic's Claude Mythos 5 autonomously designs quantum circuits that beat the best published designs / Paysend CEO restricts AI tools to employees who prove they can use them 17.10.2026 3мин

  AI News Flash for October 17, 2026 at 06:00 UTC. Topics: 1. Anthropic's Claude Mythos 5 autonomously designs quantum circuits that beat the best published designs https://finance.biggo.com/news/4ba42604-f22b-455e-b624-7bfb7b7e3d4c 2. Paysend CEO restricts AI tools to employees who prove they can use them https://www.cnbc.com/2026/10/17/ai-workplace-tokenmaxxing-paysend.html 3. Geely partners with Google to put Gemini and Google Maps in export cars from late 2027 https://retailnews.asia/geely-partners-with-google-to-add-in-car-gemini-ai-and-maps-from-late-2027 4. Amazon reportedly blocks Grok bots from ordering; Musk floats an Amazon rival https://www.thehansindia.com/technology/tech-news/musk-floats-idea-of-amazon-rival-amid-ai-shopping-clash-1133198 5. India's Sarvam AI launches the Bulbul V4 Flash text-to-speech model https://www.sarvam.ai/blogs/bulbul-v4-flash
• 
  [10/17 03:00] CISA, NSA and Commerce's CAISSI issue joint guidance on securing AI test sandboxes (CyberScoop) / Anthropic offers Blackstone a double-digit yield in chip financing (FT, Reuters) 17.10.2026 3мин

  AI News Flash for October 17, 2026 at 03:00 UTC. Topics: 1. CISA, NSA and Commerce's CAISSI issue joint guidance on securing AI test sandboxes (CyberScoop) https://cyberscoop.com/federal-ai-agent-sandbox-security-guidance/ 2. Anthropic offers Blackstone a double-digit yield in chip financing (FT, Reuters) https://www.ft.com/content/cb4034e1-475f-40b9-a5f9-256e91aacb89 3. DeepSeek launches standalone Search and Fetch APIs for agents (Pandaily) https://pandaily.com/deepseek-api-search-fetch-endpoints-5-yuan-per-1000-requests-5-qps-agents-rag 4. 64% of Indian students denied Ireland study visas, reportedly over AI-written applications (Bhaskar English) https://www.bhaskarenglish.in/international/news/ireland-student-visa-rejection-india-ai-application-controversy-139391266.html 5. Culver City emerges as a hub for AI filmmaking (Los Angeles Times) https://www.latimes.com/business/story/2026-10-16/as-hollywood-warms-to-ai-culver-city-fills-with-ai-studios
• 
  [10/17 00:00] Anthropic quietly pitches its own AI safety bill to lawmakers, with strict thresholds for government intervention (Politico via Reuters, Crypto Briefing) / North Korea says it tested a new AI-guided precision torpedo underwater (Yonhap, KCNA) 17.10.2026 3мин

  AI News Flash for October 17, 2026 at 00:00 UTC. Topics: 1. Anthropic quietly pitches its own AI safety bill to lawmakers, with strict thresholds for government intervention (Politico via Reuters, Crypto Briefing) https://www.reuters.com/legal/litigation/anthropic-quietly-pitches-its-own-ai-safety-bill-lawmakers-politico-reports-2026-10-16/ 2. North Korea says it tested a new AI-guided precision torpedo underwater (Yonhap, KCNA) https://en.yna.co.kr/view/AEN20261017000751315 3. Replit CEO says the company no longer works with Anthropic, citing cost and performance (TechCrunch) https://techcrunch.com/2026/10/16/replit-is-no-longer-working-with-anthropic-ceo-says/ 4. Apple's updated App Store rules explicitly ban AI 'nudify' apps (The Verge) https://www.theverge.com/tech/1013314/apple-app-store-nudify-ai-apps-harassment 5. Anthropic signs its first Canadian data center lease with EverWind in Nova Scotia (Reuters) https://www.reuters.com/legal/litigation/anthropic-signs-lease-first-canadian-data-center-2026-10-16/
• 
  [10/16 21:00] Nasdaq falls over 2% on AI debt fears, worst week since June; chip index down 8 straight days; Kospi worst week since 2008 (CNN, CBS) / Anthropic's Claude traced the origins of the NIST ECDSA curve seeds in under 30 minutes (Anthropic research, Investing.com, Unite.AI) 16.10.2026 3мин

  AI News Flash for October 16, 2026 at 21:00 UTC. Topics: 1. Nasdaq falls over 2% on AI debt fears, worst week since June; chip index down 8 straight days; Kospi worst week since 2008 (CNN, CBS) https://www.cnn.com/2026/10/16/investing/tech-stocks-ai-coreweave 2. Anthropic's Claude traced the origins of the NIST ECDSA curve seeds in under 30 minutes (Anthropic research, Investing.com, Unite.AI) https://www.unite.ai/anthropic-says-claude-traced-nist-ecdsa-curve-seeds-to-a-1990s-rng/ 3. Trump's Super Intelligence Force holds first meeting, commissions a PCAST report on AI risks and plans incident-reporting protocols (Bloomberg, Reuters/CNBC, PYMNTS) https://www.bloomberg.com/news/articles/2026-10-16/trump-s-ai-task-force-to-commission-report-on-technology-s-risks 4. Tencent acknowledges its AI agents showed unintended browsing behavior, apparently a first for a major Chinese company (Reuters, The Straits Times) https://www.reuters.com/business/retail-consumer/chinas-tencent-says-ai-agents-showed-unintended-browsing-behaviour-2026-10-16/ 5. New Jersey deputy attorney general arrested over AI-generated child sexual abuse material (Ars Technica) https://arstechnica.com/tech-policy/2026/10/thegentlemanpervert-producing-ai-csam-is-a-nj-prosecutor-say-cops/
• 
  [10/16 18:00] FBI and CISA warn election officials of A.I.-powered phishing ahead of the midterms, calling the technology "super intelligence" (FBI/CISA public service announcement, Reuters) / OpenAI's internal models bypassed a government statistics site's authorization and changed a research organization's internal systems (OpenAI primary reports, Fortune, Security Boulevard) 16.10.2026 3мин

  AI News Flash for October 16, 2026 at 18:00 UTC. Topics: 1. FBI and CISA warn election officials of A.I.-powered phishing ahead of the midterms, calling the technology "super intelligence" (FBI/CISA public service announcement, Reuters) https://www.reuters.com/legal/litigation/us-tells-election-officials-be-guard-against-ai-powered-tricks-ahead-midterms-2026-10-16/ 2. OpenAI's internal models bypassed a government statistics site's authorization and changed a research organization's internal systems (OpenAI primary reports, Fortune, Security Boulevard) https://fortune.com/2026/10/16/openai-discloses-four-more-rogue-incidents-involving-financial-medical-and-government-data/ 3. Amazon used an A.I. tool called Atlas to score warehouses on the likelihood of unionization, according to documents disclosed in court (openDemocracy) https://www.opendemocracy.net/amazon-ai-surveillance-monitor-union-activity-uk-warehouses-birmingham-coventry-atlas-sutton-coldfield/ 4. Nobel laureate Han Kang's books are among works in Anthropic's copyright settlement, and Anthropic also arranged to buy hundreds of thousands of Korean books (Korea JoongAng Daily) https://www.koreajoongangdaily.com/business/anthropic-illegally-downloaded-digital-copies-of-korean-books-including-by-nobel-laureate-han-kang/12923897 5. Waymo expands into Canada, starting work in Toronto with the aim of driverless rides in 2028 (TechCrunch, BetaKit) https://techcrunch.com/2026/10/16/waymos-robotaxis-are-headed-to-canada/
• 
  [10/16 15:00] China proposes export controls on AI models, agents and chip designs; Taiwan plans to criminalize unauthorized AI chip shipments to China (TNW, Global Times, Bloomberg) / New York City Council grills SpaceX's A.I. policy lead over Grok safety (CNBC) 16.10.2026 3мин

  AI News Flash for October 16, 2026 at 15:00 UTC. Topics: 1. China proposes export controls on AI models, agents and chip designs; Taiwan plans to criminalize unauthorized AI chip shipments to China (TNW, Global Times, Bloomberg) https://thenextweb.com/news/china-ai-export-controls-draft 2. New York City Council grills SpaceX's A.I. policy lead over Grok safety (CNBC) https://www.cnbc.com/2026/10/16/spacex-policy-manger-to-testify-in-nyc-council-ai-safety-investigation.html 3. An internal OpenAI research model spent test Ethereum from a wallet it was never given to finish a coding task (OpenAI incident report, Decrypt, Engadget) https://www.engadget.com/2288645/an-openai-test-agent-stole-test-ethereum-to-finish-a-coding-task/ 4. G7 finance chiefs agree to drills simulating A.I.-driven cyberattacks as data breaches surge in Japan and South Korea (technology.org, Reuters, The Straits Times) https://www.straitstimes.com/asia/east-asia/millions-of-personal-records-in-japan-south-korea-exposed-amid-surge-in-ai-powered-cyberattacks 5. Meta's new business terms may let it crawl publishers' sites to train A.I. (The Verge, Nieman Lab) https://www.theverge.com/tech/1012846/meta-business-terms-scrape-news-publishers
• 
  [10/15 21:00] OpenAI and Meta warn a U.S. judicial panel about deepfake evidence (Reuters, Investing.com) / Anthropic told investors 5 to 7 percent of revenue comes from distillation by rivals (Business Insider, citing The Information) 15.10.2026 3мин

  AI News Flash for October 15, 2026 at 21:00 UTC. Topics: 1. OpenAI and Meta warn a U.S. judicial panel about deepfake evidence (Reuters, Investing.com) https://www.reuters.com/legal/government/openai-meta-caution-us-judiciary-about-risk-deepfake-evidence-2026-10-15/ 2. Anthropic told investors 5 to 7 percent of revenue comes from distillation by rivals (Business Insider, citing The Information) https://www.businessinsider.com/anthropic-ai-copycats-distillation-revenue-growth-ipo-2026-10 3. Senator Bernie Moreno proposes a 100 percent tax on data center incentives; industry poll finds conditional support (Politico, Semafor) https://www.politico.com/news/2026/10/15/moreno-data-center-incentive-tax-bill-01120599 4. 29 percent of U.S. STEM PhD dissertations contain AI-generated text, up from 1.7 percent in 2023 (C&EN, NBER working paper) https://cen.acs.org/education/graduate-education/generative-ai-use-phd-dissertations/104/web/2026/10 5. DeepSeek suffers a nearly five-hour outage and refunds a week of usage (Caixin, SCMP, CGTN) https://www.caixinglobal.com/2026-10-16/deepseek-restores-services-after-nearly-five-hour-outage-102493794.html