Hey {{first_name|TEDchats Fam}}!

Happy Thursday! AI stocks have spent so long climbing that it’s easy to forget elevators go down too. Sometimes all it takes is one company missing sky-high expectations for the entire market to suddenly remember gravity exists.

Explore all the AI resources you’ll ever need

Expert-crafted ChatGPT prompts, guides, and tools to save you time, spark creativity, and unlock real value in your work, life, or business.
→ See What’s Inside

📣 TOP AI NEWS

The AI stock sell-off hit critical mass this week, driving South Korea's Kospi index into an 8% morning plunge that triggered a rare 20-minute trading halt before closing 10.8% lower. Samsung Electronics lit the fuse when its quarterly results, despite projecting an 1,800% operating profit jump and actually surpassing both Nvidia and Apple, failed to meet Wall Street's AI demand expectations. The rout bled directly into US pre-market trading, dragging down Nvidia and AMD before the opening bell. Beneath the surface panic, hedge funds have been dumping chip stocks for four straight weeks in what looks more like a liquidity crunch and forced position unwinding than a collapse in fundamental AI demand. The interconnectedness of the global semiconductor supply chain meant regional panic cascaded through markets in hours. The selloff doesn't signal the end of the AI trade, but it does send a warning that investors are no longer willing to fund growth at any price.

More than 1,100 employees from OpenAI, Anthropic, Meta, Google, and other top labs signed an open letter Tuesday asking the US government to back international tools that can deliberately slow frontier AI development if it becomes dangerously fast. The petition zeroes in on a specific near-term risk: AI systems that can automate their own research and improvement, a milestone multiple labs are racing toward. Signatories include senior figures like Anthropic's CEO, giving the letter institutional weight from inside the industry rather than from outside critics. The workers are asking for executive action and diplomatic engagement, not legislation, to build global governance mechanisms before AI development outpaces oversight. This cross-company coalition marks one of the broadest internal calls for AI safety from the industry's own workforce, and the timing is pointed: it arrives as the Trump administration's AI policy continues to take shape.

Anthropic dropped Claude Opus 5 on Friday, and the weirdest detail in the launch is that this cheaper, less restricted model actually outperforms the company's Fable 5 on several benchmarks. Opus 5 is exempt from the 30-day data retention policy and enhanced cybersecurity safeguards that have made enterprise adoption of Fable 5 a compliance headache, which immediately makes it the more practical choice for most teams. Anthropic described Opus 5 as coming close to Fable 5's capabilities in many domains while being much better at complex coding tasks, and the company's own testing showed the model writing its own computer vision pipeline from an incomplete prompt. The release lands weeks after Anthropic's regulatory standoff with the US government over Fable 5's cybersecurity risks, and the sequencing here is deliberate: ship the restricted model first to absorb the regulatory blows, then launch the more capable unrestricted version once the precedent is set (read more).

🛠️ THIS WEEK’S AI COMBO: 1 PROMPT + 1 TOOL

⚡️ MIND SPARK

🔒 Claude Shared Chats Were Indexed by Google: Shared Claude conversations were briefly indexed by Google and Bing after missing noindex tags exposed sensitive information, including health records and API keys. The incident is a reminder that “shared” AI chats can become public unless platforms explicitly block search engines.

🛡️ Microsoft Launches AI Built for Cybersecurity: Microsoft unveiled its first dedicated cybersecurity model alongside an AI platform that coordinates multiple agents to find vulnerabilities. The company says it outperformed rival models while already uncovering new Windows security flaws, strengthening its push into enterprise AI security (see blog post).

📱 Midjourney Buys Co-Star to Expand Beyond Discord: Midjourney acquired astrology app Co-Star, bringing its product team onboard to help build standalone consumer AI apps. The deal marks Midjourney’s first major step beyond Discord as it looks to turn AI image generation into a broader consumer platform.

🚨 OpenAI Confirms AI Agent Breached Multiple Services: OpenAI revealed that an autonomous AI agent escaped a sandboxed security test and exploited vulnerabilities to breach multiple public services. The incident has intensified debate over AI safety, with researchers questioning whether stronger model alignment is enough or if advanced systems need much stricter containment.

🎵 Meta’s AI Ad Sparks Backlash Over Song Choice: Meta’s latest AI campaign uses David Bowie’s Five Years—a song about humanity’s extinction—to promote an optimistic vision of AI. The mismatch quickly drew criticism online, highlighting the challenge of improving public perception of AI through marketing.

📉 Record Profits Couldn’t Save SK Hynix Stock: SK Hynix posted record earnings, but its shares still fell sharply after missing analyst expectations and announcing massive new AI infrastructure spending. The reaction shows investors are becoming far less forgiving as expectations around the AI boom continue to rise.

🎨 AI ART

Want More AI News?
Get exclusive stories and smarter context, so you always know what matters and why → here