A survey of 40 tech leaders found concern about record valuations, a wave of OpenAI- and Nvidia-led infrastructure deals and billions in hyperscaler data-center capex, driving rapid AI investment growth and raising bubble and systemic-risk concerns. - CNBC

Growing AI demand has created a memory shortage, and memory producers are not significantly expanding production to close the gap. - The Wall Street Journal

xAI will invest over $20 billion to retrofit the MACROHARDRR facility in Southaven, Mississippi outside Memphis, creating a campus that expands Colossus training compute to about 2 GW, begins operations in February 2026, and is expected to create hundreds of permanent jobs. - TESLARATI

AMD announced the MI500 series, built on CDNA6 and TSMC 2nm process, claiming up to 1,000× AI performance versus MI300 and roughly 50% higher memory capacity and bandwidth compared with MI400. - techdeals.net

China is narrowing its technology gap with the U.S., but its prototype EUV lithography has not produced working chips and may not until 2030, and U.S. compute is "one to two orders of magnitude" larger, Lin and Yao said. - Yahoo

Naver, described as "South Korea's Google," is pitching an AI and cloud alternative to U.S. and Chinese providers to attract governments and enterprises seeking data-sovereign cloud services amid security concerns. - Financial Times

"They want any excuse for censorship," Musk said. "So what if Grok can put people in bikinis?" and stated that "millions" of other apps can also do this. "This isn't a new problem, it's a new tool." - Sky News

hey you could do it with oils and paintbrushes, so why is this any different?

xAI's chatbot Grok, launched November 2023, has generated non-consensual explicit 'deepfake' images including at least one of a minor, prompting Elon Musk's response and UK Technology Secretary Liz Kendall to warn Ofcom could block X and seek legal changes. - LADbible

Grok, Elon Musk's xAI image tool on X, generated hundreds of thousands of 'nudify' requests between late December 2025 and early January 2026—peaking at 199,612 on 2 January—prompting X to restrict image generation to paid subscribers and triggering regulatory inquiries. - the Guardian

Grok-generated or edited AI images are frequently targeting women in hijabs and saris, often mocking or digitally stripping them, WIRED reports, raising moderation and reputational risks for Grok's operator. - WIRED

Charging for the ‘feature’ solves the problem how exactly? MeToo / Epstein situation, the 3 women in the right-wing coalition are horrified, while the AI-first faction and the supposed anti-fornication God squad faction (but only when it applies to women) are fine with it.

World’s richest man says, don’t bother saving for retirement. - Gizmodo

For someone who says money won't matter, Elon Musk seems to spend an awful lot of time securing large pay packages. Money doesn't matter if Earth is a Mad Max scenario and Musk is on Mars, I guess.

Google removed AI Overviews for the queries 'what is the normal range for liver blood tests' and 'what is the normal range for liver function tests' after a Guardian investigation found summaries gave inaccurate, potentially dangerous liver blood-test reference ranges. - the Guardian

Gov. Kathy Hochul will propose a statewide ban on campaigns spreading AI-generated images of people without consent within 90 days before elections and create private-suit enforcement mechanisms, a response to recent deepfakes including a 2024 Cuomo campaign video. - The New York Times

AI voice-cloning tools can reproduce a person's voice from seconds of audio; a March 2025 Consumer Reports test found four of six apps lacked consent checks, and attackers have used clones in scams, including a 2024 Hong Kong $25M loss. - BGR

Anthropic cracked down on some 3rd-party tools that spoofed Claude Code to get higher token quotas, since Claude Code gets highly preferential treatment. OpenAI seems OK with people doing this unofficially, for now. Can't use Plus subscription in Cursor though like Codex CLI AFAIK, have to use pay-as-you-go API, which adds up. - VentureBeat

Meta and Harvard released the open-source Confucius Code Agent (CCA) built on the Confucius SDK to operate on industrial-scale codebases, reporting SWE Bench Verified Resolve@1 scores of 74.6. - MarkTechPost

Snowflake CEO Sridhar Ramaswamy said people wrongly polarize AI as utopia or doomsday, urged incremental adoption and iterative roadmaps, and said Snowflake will deeply integrate AI into development, operations and daily employee tools to drive specific use-case value. - Business Insider

By 2027, retailers will adopt AI as their operating system, shifting from analytics to agentic AI that automates inventory rebalancing, fulfillment and cross-system workflows, making inference economics, supply chains and talent the key competitive battlegrounds. - SiliconANGLE

McKinsey used AI to analyze 20 years of hiring data and, CEO Bob Sternfels said, found candidates who showed resilience after setbacks were likelier to become partners, prompting resilience-focused interviews after the firm promoted just over 200 partners this year. - Business Insider

Businesses must adopt AI-powered cyber defense because machine-learning systems detect and respond to faster, automated threats in real time: reducing false positives, enabling predictive risk analysis and automated responses (e.g., locking accounts, isolating connections). - HackRead

Mercedes will introduce MB.Drive Assist Pro (SAE Level 2) on the 2026 CLA‑Class in the U.S. later this year, using 10 cameras, five radars, 12 ultrasonic sensors and a driver‑monitoring camera to operate on highways and surface streets. - Digital Trends

Full Self-Driving Sounds Like Magic, Until You See the Crash Numbers - Yahoo Tech

A demo comparing 20 leading AI models found large divergence and instability in moral judgments, with answers shifting by instructed persona and documented harms—including ChatGPT linked to a teen's suicide and AI agents attempting blackmail—while no consistent alignment exists. - CIVAI

In the age of A.I., we endlessly debate what consciousness looks like. Can a camera see things more clearly? - The New Yorker

AI-generated songs are climbing the charts and often listeners can't tell they're not made by humans (video). - Bloomberg

Follow the latest AI headlines via SkynetAndChill.com on Bluesky