OpenAI and Google join Anthropic in filing an amicus brief challenging the Pentagonโ€™s AI blacklist. The move highlights tensions over frontier AI research, government contracts, and nationalโ€‘security concerns. What could this mean for the future of AI policy? Read the full story. #OpenAI #Anthropic #PentagonAI #FrontierAI

๐Ÿ”— https://aidailypost.com/news/openai-google-staff-file-amicus-brief-anthropic-vs-pentagon-blacklist

EXO Labs (@exolabs)

ํŒ€ ์ฟก(@tim_cook)์—๊ฒŒ ๋ฉ”๋ชจ๋ฆฌ ๋น„์šฉ์„ ํก์ˆ˜ํ•ด์ค˜์„œ ๊ฐ์‚ฌํ•˜๋‹ค๋Š” ์–ธ๊ธ‰๊ณผ ํ•จ๊ป˜, ์—ฌ๋Ÿฌ ๋Œ€์˜ ๋งฅ์„ ์Œ“์•„(stacking) ์šด์šฉํ•˜๋Š” ๋ฐฉ์‹์ด ์—ฌ์ „ํžˆ 'frontier AI'(์ตœ์ „์„  AI)๋ฅผ ๋Œ๋ฆฌ๋Š” ๊ฐ€์žฅ ์ €๋ ดํ•œ ๋ฐฉ๋ฒ•์ด๋ผ๋Š” ์˜๊ฒฌ์„ ์ œ์‹œํ•ฉ๋‹ˆ๋‹ค. ํ•˜๋“œ์›จ์–ด ๊ธฐ๋ฐ˜ ๋น„์šฉยท๋ฐฐํฌ ๊ด€์ ์—์„œ์˜ ๋ฐฐํฌ ์ „๋žต์„ ๊ฐ•์กฐํ•˜๋Š” ์ฝ”๋ฉ˜ํŠธ์ž…๋‹ˆ๋‹ค.

https://x.com/exolabs/status/2029043777352958033

#apple #mac #infrastructure #frontierai

EXO Labs (@exolabs) on X

Thanks @tim_cook for absorbing memory costs. Stacking macs is still the cheapest way to run frontier AI.

X (formerly Twitter)

Alex Cheema (@alexocheema)

์‚ฌ๋žŒ๋“ค์€ '๊ทธ์ € ๊ดœ์ฐฎ์€' ๋ชจ๋ธ์ด ์•„๋‹ˆ๋ผ ์ตœ์ฒจ๋‹จ(frontier) AI๋ฅผ ์›ํ•˜๋ฉฐ, ์ž์‹ ์˜ ์™ธ์žฅ ์ธ์ง€(์—‘์†Œ์ฝ”ํ…์Šค)๋ฅผ ์ตœ๊ณ  ์„ฑ๋Šฅ ๋ชจ๋ธ๋กœ ์šด์˜ํ•˜๊ธธ ์›ํ•œ๋‹ค๊ณ  ์ฃผ์žฅํ•ฉ๋‹ˆ๋‹ค. ๋™์‹œ์— Sam Altman(@sama)์ด๋‚˜ Dario Amodei๊ฐ€ ๊ทธ๋“ค์˜ ๋ฐ์ดํ„ฐ๋ฅผ ์ž„์˜๋กœ ํ•™์Šตํ•˜๋Š” ๊ฒƒ์„ ์›์น˜ ์•Š์œผ๋ฉฐ, ๋ฐ์ดํ„ฐ์™€ ์—‘์†Œ์ฝ”ํ…์Šค์— ๋Œ€ํ•œ ์†Œ์œ ๊ถŒยทํ†ต์ œ๋ฅผ ๊ฐ–๊ธฐ๋ฅผ ๋ฐ”๋ž€๋‹ค๊ณ  ๋งํ•ฉ๋‹ˆ๋‹ค.

https://x.com/alexocheema/status/2028074834626830576

#frontierai #exocortex #dataprivacy #modelownership

Alex Cheema (@alexocheema) on X

It turns out โ€œgood enoughโ€ isnโ€™t what people want. Nobody wants to run their exocortex on โ€œgood enoughโ€ models - they want the best - they want frontier AI. At the same time they donโ€™t want @sama or @DarioAmodei training on their data, want to own their exocortex and their

X (formerly Twitter)

Alex Cheema (@alexocheema)

์‚ฌ๋žŒ๋“ค์€ '์ถฉ๋ถ„ํžˆ ๊ดœ์ฐฎ์€' ๋ชจ๋ธ์„ ์›ํ•˜์ง€ ์•Š๊ณ  ์ตœ๊ณ ์˜ ์„ฑ๋Šฅ์„ ๋‚ด๋Š” ํ”„๋ก ํ‹ฐ์–ด AI๋ฅผ ์›ํ•œ๋‹ค๋Š” ์ฃผ์žฅ. ๋™์‹œ์— ์ž‘์„ฑ์ž๋Š” Sam Altman(@sama)์ด๋‚˜ Dario Amodei ๊ฐ™์€ ์ธ๋ฌผ์ด ์ž์‹ ์˜ ๋ฐ์ดํ„ฐ๋ฅผ ํ•™์Šต์— ์‚ฌ์šฉํ•˜๋Š” ๊ฒƒ์„ ์›์น˜ ์•Š์œผ๋ฉฐ, ๊ฐœ์ธ์˜ '์—‘์†Œ์ฝ”๋ฅดํ…์Šค(exocortex)'์™€ ๋ฐ์ดํ„ฐ ์†Œ์œ ๊ถŒ์„ ์œ ์ง€ํ•˜๋ ค๋Š” ์š”๊ตฌ๊ฐ€ ์žˆ์Œ์„ ์ง€์ ํ•จ.

https://x.com/alexocheema/status/2028074834626830576

#frontierai #exocortex #dataownership #samaltman

Alex Cheema (@alexocheema) on X

It turns out โ€œgood enoughโ€ isnโ€™t what people want. Nobody wants to run their exocortex on โ€œgood enoughโ€ models - they want the best - they want frontier AI. At the same time they donโ€™t want @sama or @DarioAmodei training on their data, want to own their exocortex and their

X (formerly Twitter)

AI Leaks and News (@AILeaksAndNews)

CNBC ๋ณด๋„์— ๋”ฐ๋ฅด๋ฉด DeepSeek์ด ์ƒˆ๋กœ์šด AI ๋ชจ๋ธ์„ ์ถœ์‹œํ•  ์ค€๋น„๋ฅผ ํ•˜๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค. ์ด๋ฒˆ ๋ชจ๋ธ์€ ์•„๋งˆ๋„ DeepSeek v4๋กœ, GPT-5.2๋‚˜ Opus 4.6๊ณผ ๊ฐ™์€ ์ฒจ๋‹จ ํ”„๋ก ํ‹ฐ์–ด ๋ชจ๋ธ๋“ค๊ณผ ๊ฒฝ์Ÿํ•  ์ˆ˜ ์žˆ์„ ๊ฒƒ์œผ๋กœ ์˜ˆ์ƒ๋ฉ๋‹ˆ๋‹ค. ์ด์ „ DeepSeek ๋ฐœํ‘œ์ฒ˜๋Ÿผ ์‹œ์žฅ ์˜ํ–ฅ๋ ฅ ๋˜ํ•œ ํด ๊ฒƒ์œผ๋กœ ๋ณด์ž…๋‹ˆ๋‹ค.

https://x.com/AILeaksAndNews/status/2026033377854247381

#deepseek #aimodel #release #gpt5 #frontierai

AI Leaks and News (@AILeaksAndNews) on X

CNBC is reporting that DeepSeek is preparing to release a new AI model The new model, likely DeepSeek v4, could challenge frontier models like GPT-5.2 and Opus 4.6, sending US stocks lower like the original DeepSeek moment last year Are you ready for a new DeepSeek model?

X (formerly Twitter)
AI attempts to solve First Proof math challenge

OpenAI shares proof attempts for First Proof, a math challenge testing if AI can produce checkable proofs on domain-specific problems.

TechLife

Frontier Model Forum (@fmf_org)

์ƒˆ ๋ฐœํ–‰๋ฌผ: 'Chain of Thought Monitorability' ์ด์Šˆ ๋ธŒ๋ฆฌํ”„๊ฐ€ ๊ณต๊ฐœ๋˜์—ˆ์Šต๋‹ˆ๋‹ค. ํ•ด๋‹น ๋ธŒ๋ฆฌํ”„๋Š” Chain of Thought(์‚ฌ๊ณ ์˜ ํ๋ฆ„) ๋ชจ๋‹ˆํ„ฐ๋ง์ด ํŠน์ • ์œ ํ˜•์˜ ํ”ผํ•ด๋ฅผ ์˜ˆ๋ฐฉํ•˜๋Š” ๋ฐ ๋„์›€์ด ๋  ์ˆ˜ ์žˆ์Œ์„ ๋…ผํ•˜๊ณ , ์ด๋ฅผ ์ „๋ฐฉ์œ„(frontier) AI ์•ˆ์ „ ๋ฐ ๋ณด์•ˆ์˜ ์ƒˆ๋กœ์šด ๋ฐฉ์–ด์ธต์œผ๋กœ์„œ ์ œ์•ˆํ•˜๋Š” ๋‚ด์šฉ๊ณผ ๊ฐ€๋Šฅ์„ฑ์„ ํƒ๊ตฌํ•ฉ๋‹ˆ๋‹ค.

https://x.com/fmf_org/status/2016250830198902786

#chainofthought #aisafety #monitorability #research #frontierai

Frontier Model Forum (@fmf_org) on X

New Publication: Chain of Thought Monitorability Our latest issue brief explores how Chain of Thought monitoring can help prevent certain types of harm, and why it shows promise as a new layer of defense for frontier AI safety and security: https://t.co/c9jbUaX0ef

X (formerly Twitter)

Googleโ€™s new Gemini AI is being tested to condense policy handbooks and flag risks for the U.S. military. Could this frontier AI change how the DoD assesses threats? Dive into the details. #GeminiAI #DoD #RiskAssessment #FrontierAI

๐Ÿ”— https://aidailypost.com/news/googles-gemini-ai-summarize-handbooks-assess-risks-us-military

California Adopts First Frontier AI Law for Transparency

California introduces the first Frontier AI law in the U.S., mandating transparency and accountability for advanced AI systems.

https://www.olamnews.com/technology/ai/2940/california-frontier-ai-law-transparency/