Skip to content

#其他

0 today

Jul 15Wednesday

Hugging Face Blog

Thinking Machines releases Inkling: a 1T-param, natively multimodal open model

Inkling is an open ~1T-param model that natively accepts image, audio, and text inputs with a 1M context window. Trained on 45T multimodal tokens, it uses a MoE architecture with 975B total and 41B active parameters. It includes MTP speculative decoding layers for faster inference and ships in BF16 and NVFP4 variants. Hugging Face provides day-0 support in transformers, SGLang, vLLM, and llama.cpp, covering agentic coding, multimodal vision, and audio tasks.

Why it matters: Thinking Machines released Inkling, a 975B-parameter MoE with 41B active, native text, image and audio support, a 1M-token context window, and 45 trillion multimodal training tokens.

Jul 9Thursday

OpenAI News

OpenAI turns its bio bug bounty into an ongoing program, doubling rewards to $50K starting with GPT-5.6

OpenAI is turning its GPT-5.5 Bio Bug Bounty into an ongoing private program, now called the OpenAI Bio Bounty Program. The focus stays on universal jailbreaks that beat its biosafety challenges. Rewards jump from $25,000 to $50,000 for both GPT-5.6 and GPT-5.5, with smaller payouts possible for partial wins. GPT-5.5 testing ends July 27, 2026; after that only GPT-5.6 is in scope. Applicants need a ChatGPT account, must sign an NDA, and past GPT-5.5 applicants don't need to reapply.

Why it matters: OpenAI upgraded its bio-safety bounty from a one-off to a permanent program with doubled rewards and clearer rules — a substantive safety-mechanism update.

Jun 25Thursday

OpenAI News

OpenAI publishes economic research paper on how Codex is reshaping work

OpenAI released an economic research paper on June 25, using internal and external usage data to track Codex adoption over the past year. By May 2026, 80.6% of sampled individual users had run at least one Codex task estimated to exceed 30 minutes of human work, and 25.6% had run tasks exceeding eight hours. Inside OpenAI, Codex now accounts for 99.8% of weekly output tokens; Legal and Recruiting switched their primary AI tool from ChatGPT to Codex around April 2026. Non-developer users grew fastest—137x for individuals, 189x for organizations. The paper does not disclose Codex pricing or external enterprise conversion rates.

Why it matters: OpenAI used a year of its own and external usage data to measure how far Codex-style tools reach into personal tasks, with 80.6% of users running jobs expected to take over 30 minutes.

Google Research Blog

How reasoning unlocks parametric knowledge in LLMs

Google Research shows that letting models think before answering sharply improves their ability to recall facts from training data. On Natural Questions, Gemini 2.5 Pro jumps from ~40% accuracy without reasoning to over 70% with it. The gain comes from the model connecting fuzzy memories into verifiable chains, not from external retrieval. The reasoning traces often include self-questioning and fact-checking steps. The post only covers QA tasks so far.

Why it matters: Google's test on Natural Questions shows reasoning first lifts Gemini 2.5 Pro accuracy from about 40% to over 70%, turning fuzzy recall into checkable fact chains.

Jun 22Monday

OpenAI News

OpenAI launches Patch the Planet to help open source maintainers patch bugs, not just find them

OpenAI's Daybreak initiative partners with Trail of Bits to pair GPT‑5.5‑Cyber and Codex Security with human review, finding and patching vulnerabilities across 19 critical open source projects including cURL, Go, and Python. Security engineers filter false positives and develop patches before handing off to maintainers. The initial sprint found hundreds of issues, merged dozens of patches, and built reusable fuzzing and variant-analysis pipelines.

Why it matters: OpenAI and Trail of Bits had security engineers use two models to scan 19 critical open source projects including cURL, Go and Python and submit patches, running the full detect-to-fix loop.

OpenAI News

Samsung Electronics rolls out ChatGPT and Codex to employees in one of OpenAI's largest enterprise deals

Samsung Electronics is deploying ChatGPT Enterprise and Codex to all employees in Korea and its DX division worldwide, covering R&D, manufacturing, marketing, and more. OpenAI calls it one of its largest enterprise launches ever. Codex now has over 5 million weekly active users; weekly actives in Korea grew nearly 800% since Feb 1, 2026. The post does not disclose deal value or rollout timeline.

Why it matters: Samsung deploying ChatGPT Enterprise and Codex across Korean staff and global DX units is one of OpenAI's larger enterprise rollouts.

Jun 18Thursday

OpenAI News

OpenAI o3 Deep Research reanalyzed 376 unsolved pediatric cases and surfaced leads for 18 rare-disease diagnoses

Boston Children’s, Harvard, and OpenAI used o3 Deep Research to reanalyze 376 previously unsolved pediatric rare-disease cases. The model proposed evidence-linked hypotheses; after expert review and lab confirmation, physicians established 18 new diagnoses—an additional yield of 4.8%. The model never made clinical decisions. All confirmed diagnoses went through CLIA-certified lab validation. The study appears in NEJM AI and the authors note it is a retrospective analysis, not yet a routine clinical tool.

Why it matters: Boston Children's Hospital, Harvard and OpenAI reran 376 undiagnosed pediatric rare disease cases through o3, adding 18 diagnoses on top of what specialists had already concluded, with confirmation from a CLIA-certified lab.

Jun 17Wednesday

Hugging Face Blog

Z.AI releases GLM-5.2: first open-source model with solid 1M-token context, built for long-horizon coding tasks

Z.AI open-sourced GLM-5.2, a model built for long-horizon coding tasks. It delivers a genuinely usable 1M-token context—not just accepting more tokens, but maintaining quality across long agent trajectories. IndexShare reuses one indexer across every four sparse attention layers, cutting per-token FLOPs by 2.9× at 1M context; MTP acceptance length improved by up to 20%. On FrontierSWE it beats GPT-5.5 by 1%, and on PostTrainBench it outranks both GPT-5.5 and Opus 4.7, placing second. It's the top open-source model across all three long-horizon coding benchmarks. MIT license, no regional restrictions.

Why it matters: Z.AI's open-source GLM-5.2 pushes context to 1M tokens, holds output quality through long agent runs, and IndexShare cuts compute per token to about a third at equal length.

Jun 16Tuesday

OpenAI News

OpenAI simulates real-world deployment to catch undesired model behavior before release

OpenAI replays recent real conversations through a candidate model before release, then checks for new undesired behaviors. Across GPT‑5‑Thinking deployments, this Deployment Simulation gave more accurate frequency estimates than traditional evals, surfaced novel misalignment, and reduced the chance models could tell they were being tested. It also works for agentic rollouts with tool use. The method can’t catch behaviors rarer than 1 in 200,000 messages.

Why it matters: OpenAI replayed real conversations to have a new model regenerate replies before GPT-5-Thinking shipped, estimating bad-behavior rates and surfacing alignment issues it had not seen before.

Jun 10Wednesday

OpenAI News

OpenAI banned PRC-linked ChatGPT accounts running covert influence ops on US AI debates

OpenAI published a threat report on June 10 detailing two clusters of ChatGPT accounts likely originating from China, both banned for covert influence operations. One cluster, named 'Data Center Bandwagon,' generated posts claiming AI data centers were raising household electricity prices. The other, 'Tech and Tariffs,' criticized US tariffs as tech competition tactics and instructed outputs to mention only President Trump, not Xi Jinping. That second cluster also spread false claims of a ChatGPT user data breach, which OpenAI calls entirely fabricated. OpenAI found no evidence the operations shifted public opinion, but sees them as testing narratives against US AI infrastructure. The post does not disclose account counts, target platforms, or reach metrics.

Why it matters: OpenAI's threat report names two China-linked account clusters posting about data center power bills and tech tariffs, a concrete case of covert influence in AI policy talk.

NVIDIA Blog

NVIDIA confidential computing will help Apple expand Private Cloud Compute

Apple is bringing NVIDIA's confidential computing into Private Cloud Compute, running AI inference inside encrypted GPU environments. The setup uses H100 GPUs and Hopper architecture with hardware-level trusted execution environments, so data stays encrypted during processing and even the cloud provider can't access it. Apple previously ran private cloud inference only on its own silicon; this deal signals a shift of some workloads to NVIDIA while keeping the same security isolation. The post doesn't give a launch date or scale numbers, but confirms deployment will start in Apple's own data centers.

Why it matters: Apple moved part of its Private Cloud Compute inference load from its own silicon to Nvidia H100s, keeping data encrypted through hardware-level trusted execution, so its compute base now spans general GPUs.

Jun 1Monday

OpenAI News

OpenAI banned a likely PRC-origin cluster using ChatGPT to generate anti-US-data-center social media content

OpenAI's June threat report details a banned cluster of ChatGPT accounts likely originating in China. The operators used Simplified Chinese prompts to generate English posts and images on X, posing as ordinary Americans and claiming data centers and AI are driving up electricity costs for households. They also used ChatGPT for image editing, automation scripts, and harassing overseas dissidents. An internal work report they uploaded outlined tactics for building credible personas on Facebook and evading platform detection.

Why it matters: OpenAI banned Chinese accounts that used Simplified Chinese prompts to generate English posts impersonating Americans and push the data-center electricity-cost angle on X.

OpenAI News

OpenAI bans PRC-linked accounts using ChatGPT to generate comments on US tech policy and tariffs

OpenAI's June 2026 threat report details a banned cluster of ChatGPT accounts likely originating in China. The operators used Simplified Chinese prompts and VPNs to generate English comments and political cartoons criticizing US tariffs, rare earths, AI, and 5G policy. They instructed the model to depict only Trump, not Xi Jinping or China. The same cluster produced Chinese-language water army content attacking the US and Israel, amplifying anti-Jewish tropes, and harassing dissidents. OpenAI also linked these accounts to a separate X network that falsely claimed ChatGPT user data was compromised. The post does not disclose the exact number of banned accounts or the operators' specific institutional affiliation.

Why it matters: OpenAI's threat report says it banned accounts likely from China that used Simplified Chinese prompts to mass-produce English comments and cartoons aimed at US tariff and tech policy debates.

Apr 25Saturday

OpenAI News

OpenAI's jobs framework maps 921 occupations into four transition paths

OpenAI's AI Jobs Transition Framework sorts ~148M US jobs across 921 occupations into four paths: 18% at higher automation risk, 24% likely to reorganize, 12% could grow with AI-driven demand, and 46% face less immediate change. It goes beyond task exposure by asking whether a person remains central to delivery and whether lower costs expand demand. ChatGPT usage is roughly 3x higher in the most at-risk occupations, but recent unemployment shifts don't line up neatly with technical exposure. The report stresses that capability doesn't equal instant displacement—employers still need to rework workflows and weigh costs.

Why it matters: OpenAI sorted about 148 million US jobs across 921 occupations into four automation-risk tiers, with 18% at higher risk, based on how standardized the information is and demand elasticity.

Apr 16Thursday

OpenAI News

Codex for (almost) everything

OpenAI published a post titled "Codex for (almost) everything." The provided content has no body text, so the only confirmed facts are the mention of Codex and the phrase "almost everything," which is not enough to verify features, timing, or scope.

Why it matters: OpenAI updates Codex for more than 3 million weekly developers, letting it operate other Mac apps in the background and run several tasks in parallel, extending it from writing code to executing desktop actions.

Feb 1Sunday

OpenAI News

OpenAI banned an account linked to Chinese law enforcement that used ChatGPT to plan a covert influence operation targeting Japan's prime minister

OpenAI banned a ChatGPT account tied to a person linked to Chinese law enforcement who used the model to plan a covert influence operation against Japan's prime minister; the model refused that request. The user also asked the model to polish regular status reports on what they called cyber special operations, which mentioned locally deployed models including DeepSeek and Qwen and hundreds of personnel. OpenAI's open-source investigation found cross-platform activity matching the reports, including hashtags the Chinese text calls “右翼共生者” (roughly, right-wing symbionts), though its own models did not generate that content.

Why it matters: OpenAI says it banned an account tied to Chinese law enforcement that used ChatGPT to plan a covert influence operation against Japan's prime minister, a request the model refused.

OpenAI News

OpenAI banned Russia-linked ChatGPT accounts used to mass-produce multilingual influence content

OpenAI disclosed 'Fish Food' on Feb 1, banning ChatGPT accounts tied to the Russia-linked Rybar network. The accounts used Russian prompts to mass-generate content in English, Spanish, and other languages, posted via unaffiliated Telegram and X accounts. One batch of 7 tweets hit over 150K views for the top post from an account with 600K+ followers. The actor also drafted Africa election-interference proposals, with the priciest plan budgeted at $600K a year.

Why it matters: OpenAI says it banned ChatGPT accounts tied to the Rybar network that used Russian-language prompts to generate multilingual content and post it across social platforms, forming an AI content farm pipeline.

OpenAI News

OpenAI bans Cambodia-based accounts using ChatGPT to impersonate lawyers and the FBI in recovery scams

OpenAI's February safety report details 'Operation False Witness': a cluster very likely based in Cambodia used ChatGPT to create content for at least six fake law firms and to impersonate the FBI's IC3 unit. They lured fraud victims via social media ads, then used ChatGPT to draft lawyer-style messages on Telegram, demanding upfront crypto payments—such as a 15% service fee—before any recovery. The scammers also generated fake New York State Bar membership cards and bogus confidentiality agreements. OpenAI says individual victims may have lost thousands of dollars but cannot independently verify the amounts. The FBI and at least one impersonated firm have issued public alerts.

Why it matters: OpenAI says Cambodia-based scam groups used ChatGPT to mass-produce fake law firm promotions, forge lawyer identities and impersonate the FBI's IC3 site to run recovery scams against fraud victims.

Oct 1, 2025Wednesday

OpenAI News

OpenAI banned accounts using ChatGPT to build Russian-speaking malware tooling

OpenAI's October threat report details banned ChatGPT accounts linked to Russian-speaking criminal groups. The operators used the model to prototype malware loaders, credential stealers, RAT components, and evasion layers. Direct malicious requests were refused, so they elicited building-block code and assembled it into attack workflows offline. OpenAI found no evidence the models provided novel capabilities and shared indicators with industry partners.

Why it matters: A concrete case of how a Russian-speaking crime group split malware into small pieces to slip past model safety filters.

OpenAI News

OpenAI bans accounts using ChatGPT for phishing and scripting support tied to PRC intelligence requirements

OpenAI's October threat report details banned ChatGPT accounts whose activity overlapped with publicly tracked groups UNK_DROPPITCH and UTA0388. The actors used the model to draft phishing emails in Chinese, English, and Japanese, and to assist Go and PowerShell malware development—including encrypted C2 and process enumeration. The model introduced no novel offensive capabilities; the operators sought incremental speed and localization, yet left giveaway errors like implausible contact details in email signatures. They also explored DeepSeek for automating mass phishing, though OpenAI cannot confirm whether that work proceeded.

Why it matters: OpenAI ties banned accounts to UNK_DROPPITCH and UTA0388, which used models to write localized phishing emails aimed at Taiwan semiconductor targets.

OpenAI News

OpenAI banned a covert PRC-linked influence network using its models to mass-produce posts on the South China Sea and Hong Kong

OpenAI's October 2025 report details a covert influence network, dubbed 'Nine-emdash Line,' that used ChatGPT to generate English and Cantonese social media posts. The content attacked Vietnam's environmental record in the South China Sea, smeared Philippine President Marcos, and discredited Hong Kong pro-democracy figures. The operators also used the models for recon—finding niche forums and generating Tibetan names for fake accounts—and sought advice on launching TikTok challenges to hijack trending hashtags. OpenAI banned the accounts; many X accounts were suspended. No technical link to the known 'Spamouflage' operation was found.

Why it matters: OpenAI's October 2025 report bans accounts using ChatGPT to mass-produce English and Cantonese posts targeting South China Sea records, Marcos and Hong Kong democrats.

Jun 1, 2025Sunday

OpenAI News

OpenAI banned accounts using ChatGPT for social engineering and fake personas

OpenAI's June threat report details 'VAGue Focus': a cluster of ChatGPT accounts used to generate social media posts, translate phishing-style messages, and pose as Europe- and Turkey-based consultancies for intelligence collection. The accounts operated mostly during mainland China business hours with Chinese prompts. They impersonated 'Focus Lens News,' 'Visionary Advisory Group,' and others, cold-messaging journalists and researchers on X, and claimed to offer $2,000 per hour for interviews. Public engagement was near zero; the one account with 17K followers was likely compromised and repurposed. OpenAI banned the network. The post does not identify who was behind it.

Why it matters: OpenAI's June threat report names a campaign that used ChatGPT to mass-produce social posts, translate phishing emails and pose as consultants.

OpenAI News

OpenAI bans accounts tied to AI-generated fake remote-job applications

OpenAI's June threat intel report details banned ChatGPT accounts linked to fraudulent remote-job campaigns. Operators used the models to mass-generate tailored résumés, answer interview questions, and research tools like Tailscale and OBS to mask remote laptop locations. The behavior matches publicly attributed North Korean IT worker schemes, with some collaborators possibly receiving corporate laptops inside the US. The post doesn't quantify how many companies were affected or the financial damage, but OpenAI shared the findings with industry peers and authorities.

Why it matters: OpenAI's threat intelligence report says it banned accounts used for remote-job fraud, with details on each step from generating resumes to real-time help during video interviews.

OpenAI News

OpenAI banned accounts using ChatGPT to mass-produce Philippine political comments

OpenAI's June report details a takedown named 'High Five.' A Philippine marketing firm, Comm&Sense Inc, used ChatGPT to mass-produce short English and Taglish comments praising President Marcos and attacking VP Duterte on TikTok and Facebook. The workflow included analyzing political posts, generating sub-10-word comments, and running five coordinated TikTok channels. OpenAI banned the accounts; the actor tried to return multiple times. Thousands of comments were posted, but none got more than single-digit engagement.

Why it matters: In its monthly safety report, OpenAI names Philippine marketing firm Comm&Sense, which used ChatGPT to analyze posts about Marcos and Duterte, then mass-generated short comments and flooded TikTok and Facebook comment sections.

Feb 1, 2025Saturday

OpenAI News

OpenAI bans China-linked accounts that used ChatGPT to plant anti-US articles in Latin American media

OpenAI banned ChatGPT accounts likely tied to China that generated English posts attacking dissident Cai Xia and Spanish-language articles criticizing the US. The Spanish articles appeared on news sites in Peru, Mexico, and Ecuador, some labeled as sponsored content, with bylines pointing to a Jilin-based company. OpenAI says this is the first observed case of a China-origin influence operation successfully placing long-form articles in Latin American mainstream media, rating it Category 4 on the Breakout Scale. Social media engagement was minimal; the paid articles may have reached a wider audience.

Why it matters: OpenAI details a covert influence operation that used ChatGPT to post fake foreign-user comments on X against Cai Xia and place Spanish-language articles in Latin American media.

OpenAI News

OpenAI banned a Cambodia-based cluster using ChatGPT for pig-butchering scams

OpenAI banned a cluster of ChatGPT accounts originating in Cambodia that were used to translate and generate romance-investment scam conversations in Japanese, Chinese, and English. The scammers targeted men over 40 on Facebook, X, and Instagram using stolen influencer photos, then moved chats to LINE or WhatsApp within days. OpenAI reconstructed a six-step workflow from public engagement to fraudulent investment, noting the actors provided the model with detailed fake personas and used it mainly for translation and flirty replies.

Why it matters: OpenAI banned Cambodia-based ChatGPT accounts that wrote scam scripts in Chinese, Japanese and English targeting men over 40, from public chat to private messaging apps.

OpenAI News

OpenAI banned China-linked accounts using ChatGPT for surveillance-tool pitches and document analysis

OpenAI disclosed in Feb 2025 that it banned a cluster of ChatGPT accounts likely from China, dubbed “Peer Review.” The operators used the models to analyze English document screenshots, draft sales pitches for a “Qianyue Overseas Public Opinion AI Assistant,” and debug related code. The tool claimed to scrape X, Facebook, and other platforms to spot China-related protest calls and report them. Code debugging primarily invoked Meta’s Llama 3.1 8B, with references to Alibaba’s Qwen and an unspecified DeepSeek model. OpenAI found no evidence the generated content was posted publicly and said impact assessment requires input from other model providers.

Why it matters: OpenAI banned a China-linked network it calls Peer Review that used its models to translate and analyze notices about Uyghur human rights protests.