Microsoft expands its bug bounty program so that any critical vulnerability impacting its online services, including vulnerabilities in third-party code, is eligible for awards.
confirmed 1 article 2025-12-13
Microsoft bug bounty program third-party code online services
Microsoft expands its bug bounty program so that any critical vulnerability, including in third-party code, impacting its online services is eligible for awards
SiliconANGLE · 2025-12-13 · #24
OpenEvidence is reported to have fewer than 1 million users.
confirmed 1 article 2025-12-13
OpenEvidence <1M users
Source: OpenEvidence, which offers an AI chatbot for doctors, is raising $250M at a $12B valuation, up from $6B in Oct., and has $150M in annualized ad revenue
The Information · 2025-12-13 · #28
Tinker offers OpenAI API–compatible inference for easier integration with existing tools and workflows.
confirmed 1 article 2025-12-13
Thinking Machines Lab OpenAI API
Mira Murati's Thinking Machines Lab makes Tinker, its API for fine-tuning language models, generally available, adds support for Kimi K2 Thinking, and more
Thinking Machines Lab · 2025-12-13 · #32
Substack says it reverted the app-required feature after complaints.
confirmed 1 article 2025-12-12
Substack
[Thread] Substack quietly added a feature that forced subscribers to download the app to read full newsletters, but says it reverted the change after complaints
@gergelyorosz · 2025-12-12 · #30
xAI will deploy Grok across 5,000+ public schools in El Salvador, reaching over 1 million students over the next two years.
confirmed 1 article 2025-12-12
xAI El Salvador Grok 5,000+ public schools 1M+ students two-year timeframe
xAI partners with El Salvador to deploy Grok across 5,000+ public schools, reaching 1M+ students in an “AI-powered education program” over the next two years
The Guardian · 2025-12-12 · #36
OpenAI celebrates its 10-year anniversary.
confirmed 1 article 2025-12-12
OpenAI
On OpenAI's 10-year anniversary, Sam Altman reflects on “a decade of breakthroughs, learnings, and the path toward AGI that benefits all of humanity”
OpenAI · 2025-12-12 · #37
Nikita Bier, X's head of product, says he wants to rebuild X's recommendation algorithm to lure journalists back to the platform.
confirmed 1 article 2025-12-12
Nikita Bier X recommendation algorithm
An interview with Nikita Bier, who wants to lure journalists back to X with a rebuilt recommendation algorithm with no keyword suppression or manual downranking
Sources · 2025-12-12 · #9
OpenAI's Head of Applications Fidji Simo says she expects ChatGPT's 'adult mode' to launch in Q1 2026.
confirmed 1 article 2025-12-12
Fidji Simo OpenAI ChatGPT adult mode
OpenAI's Fidji Simo says she expects ChatGPT's “adult mode” to launch in Q1 2026 and that OpenAI wants to get better at age prediction before adding the feature
The Verge · 2025-12-12 · #42
Reports indicate GPT-5.2 Pro (High) achieved approximately 54.2% on ARC-AGI-2.
confirmed 1 article 2025-12-12
GPT-5.2 Pro ~54.2% on ARC-AGI-2
OpenAI says GPT‑5.2 Thinking beats or ties industry professionals on 70.9% of GDPval knowledge work tasks, delivering outputs at >11x the speed and <1% the cost
OpenAI · 2025-12-12 · #40
Substack rolled out a feature that required subscribers to download the app to read full newsletters.
confirmed 1 article 2025-12-12
Substack
[Thread] Substack quietly added a feature that forced subscribers to download the app to read full newsletters, but says it reverted the change after complaints
@gergelyorosz · 2025-12-12 · #30
The Kirin 9030 chip is used in Huawei's Mate 80 Pro Max smartphone.
confirmed 1 article 2025-12-12
Kirin 9030 Mate 80 Pro Max
TechInsights: Huawei's Kirin 9030 chip, used in its Mate 80 Pro Max phone, is China's most advanced chip to date and was produced with SMIC's updated 7nm tech
Bloomberg · 2025-12-12 · #20
Intel has tested chipmaking tools in 2025 supplied by California-based ACM Research, a toolmaker with deep China ties, according to sources.
rumored 1 article 2025-12-12
Intel ACM Research
Sources: Intel has tested chipmaking tools in 2025 from California-based ACM Research, which has deep China ties; US sanctions targeted two ACM overseas units
Reuters · 2025-12-12 · #15
YouTube launches an option allowing US-based creators to receive payouts in PayPal's PYUSD stablecoin.
confirmed 1 article 2025-12-12
YouTube PayPal PYUSD
YouTube launches an option for US-based creators to receive payouts in PayPal's stablecoin, PYUSD
Fortune · 2025-12-12 · #16
Google (Alphabet) will build three subsea cables to serve Papua New Guinea.
confirmed 1 article 2025-12-12
Google Alphabet Papua New Guinea three subsea cables
Papua New Guinea says Google will build three subsea cables for the country, a $120M effort funded by Australia, amid China boosting its influence in the region
Reuters · 2025-12-12 · #19
Google launches Disco, a Gemini 3-powered experiment that suggests and creates GenTabs (web apps) based on users' open browser tabs and Gemini chat history.
confirmed 1 article 2025-12-12
Google Disco GenTabs Gemini 3
Google launches Disco, a Gemini 3-powered experiment that suggests and creates GenTabs, web apps based on the user's open browser tabs and Gemini chat history
TechCrunch · 2025-12-12 · #43
Huawei's Kirin 9030 chip is presented as a new flagship chip offering from Huawei.
confirmed 1 article 2025-12-12
Huawei Kirin 9030
TechInsights: Huawei's Kirin 9030 chip, used in its Mate 80 Pro Max phone, is China's most advanced chip to date and was produced with SMIC's updated 7nm tech
Bloomberg · 2025-12-12 · #20
SMIC produced the Kirin 9030 using its updated 7nm manufacturing process technology.
confirmed 1 article 2025-12-12
SMIC Kirin 9030 updated 7nm tech
TechInsights: Huawei's Kirin 9030 chip, used in its Mate 80 Pro Max phone, is China's most advanced chip to date and was produced with SMIC's updated 7nm tech
Bloomberg · 2025-12-12 · #20
The GPT-5.2 release is framed as a rapid response to competitive pressure from Google's Gemini 3 and related advances, following an internal 'code red'.
confirmed 1 article 2025-12-12
OpenAI GPT-5.2 Google Gemini 3
OpenAI says GPT‑5.2 Thinking beats or ties industry professionals on 70.9% of GDPval knowledge work tasks, delivering outputs at >11x the speed and <1% the cost
OpenAI · 2025-12-12 · #40
OpenAI says GPT-5.2 Thinking beats or ties industry professionals on 70.9% of GDPval 'well-specified' knowledge-work tasks.
confirmed 1 article 2025-12-12
OpenAI GPT-5.2 Thinking GDPval 70.9% wins/ties vs industry professionals
OpenAI says GPT‑5.2 Thinking beats or ties industry professionals on 70.9% of GDPval knowledge work tasks, delivering outputs at >11x the speed and <1% the cost
OpenAI · 2025-12-12 · #40
OpenAI claims GPT-5.2 delivers outputs at over 11× the speed and at under 1% of the cost compared to baseline human/economic costs for the evaluated tasks.
confirmed 1 article 2025-12-12
OpenAI >11x speed <1% cost GPT-5.2
OpenAI says GPT‑5.2 Thinking beats or ties industry professionals on 70.9% of GDPval knowledge work tasks, delivering outputs at >11x the speed and <1% the cost
OpenAI · 2025-12-12 · #40
OpenAI offers GPT-5.2 in multiple variants (Instant, Thinking, Pro) targeting different latency, capability, and price/performance tradeoffs.
confirmed 1 article 2025-12-12
OpenAI GPT-5.2 Instant GPT-5.2 Thinking GPT-5.2 Pro
OpenAI says GPT‑5.2 Thinking beats or ties industry professionals on 70.9% of GDPval knowledge work tasks, delivering outputs at >11x the speed and <1% the cost
OpenAI · 2025-12-12 · #40
Sam Altman says he expects OpenAI to exit its internal 'code red' state by January following the GPT-5.2 launch.
confirmed 1 article 2025-12-12
Sam Altman exit 'code red' by January GPT-5.2 launch
OpenAI says GPT‑5.2 Thinking beats or ties industry professionals on 70.9% of GDPval knowledge work tasks, delivering outputs at >11x the speed and <1% the cost
OpenAI · 2025-12-12 · #40
GPT-5.2 models support up to 128K max output tokens, the same as GPT-5 and GPT-5.1.
confirmed 1 article 2025-12-12
GPT-5.2 128K tokens
GPT-5.2 models have a 400K context window and 128K max output tokens, like GPT-5 and 5.1, but have an August 31, 2025, knowledge cutoff, vs. September 30, 2024
Simon Willison's Newsletter · 2025-12-12 · #34
GPT-5.2 models have a knowledge cutoff of August 31, 2025, compared to GPT-5/GPT-5.1's September 30, 2024.
confirmed 1 article 2025-12-12
GPT-5.2 2025-08-31
GPT-5.2 models have a 400K context window and 128K max output tokens, like GPT-5 and 5.1, but have an August 31, 2025, knowledge cutoff, vs. September 30, 2024
Simon Willison's Newsletter · 2025-12-12 · #34
OpenAI launches GPT-5.2 as a new flagship model (available in Instant, Thinking, and Pro variants) and rolls it out to users and the API.
confirmed 1 article 2025-12-12
OpenAI GPT-5.2
OpenAI says GPT‑5.2 Thinking beats or ties industry professionals on 70.9% of GDPval knowledge work tasks, delivering outputs at >11x the speed and <1% the cost
OpenAI · 2025-12-12 · #40
Competitors Beehiiv, Ghost, and Buttondown continue to send full newsletter content in emails by default.
confirmed 1 article 2025-12-12
Beehiiv Ghost Buttondown
[Thread] Substack quietly added a feature that forced subscribers to download the app to read full newsletters, but says it reverted the change after complaints
@gergelyorosz · 2025-12-12 · #30
Sources report Nvidia told its Chinese clients it is evaluating adding production capacity for its H200 chips after orders exceeded its current output level.
rumored 1 article 2025-12-12
Nvidia H200 chips Chinese clients
Sources: Nvidia tells its Chinese clients that it is evaluating adding production capacity for its H200 chips after orders exceeded its current output level
Reuters · 2025-12-12 · #31
Taiwan opens its largest AI supercomputing data center.
confirmed 1 article 2025-12-12
Taiwan AI supercomputing data center
Taiwan opens its largest AI supercomputing data center, with Nvidia's Blackwell chips, a major effort in its push for sovereign AI and chip industry innovation
Nikkei Asia · 2025-12-12 · #33
Sandfall Interactive developed the game Clair Obscur: Expedition 33.
confirmed 1 article 2025-12-12
Sandfall Interactive Clair Obscur: Expedition 33
A look at France-based indie game studio Sandfall Interactive, whose critically acclaimed Clair Obscur: Expedition 33 had a sub-$10M budget and sold 5M+ copies
New York Times · 2025-12-12 · #35
Substack changed email behavior so tapping 'view entire message' required the app and full content was no longer sent in the email.
confirmed 1 article 2025-12-12
Substack
[Thread] Substack quietly added a feature that forced subscribers to download the app to read full newsletters, but says it reverted the change after complaints
@gergelyorosz · 2025-12-12 · #30