科技 · Section
The Decoder
AI 新闻与分析，英文全文
- OpenAIs GPT-6 Cyber can both break into systems and propose how to fix themOpenAI has introduced GPT-6 Cyber, an AI model trained specifically to hunt for security vulnerabilities in software. OpenAI says the model can find flaws, build working exploits, and suggest patches. It's also supposed to stay within its a
TDThe Decoder - Google's Gemini Omni 1.2 Flash stretches AI video clips to 15 seconds and cuts pricing by 30 percentGoogle Deepmind is making its Gemini Omni 1.2 Flash video-generation model generally available. The updated model creates clips up to 15 seconds long, supports conversational editing, and costs 30 percent less than its predecessor. Google s
TDThe Decoder - Baidu releases ERNIE 5.5 Flash Base with 280 billion parameters as an open-source foundation modelBaidu's AI development is still alive, but most of the work is on you. The Chinese company has released ERNIE-5.5-Flash-Base , an open foundation model that ships deliberately without fine-tuning or post-training. Developers are expected to
TDThe Decoder - No AI agent scores above 12 percent on SemiAnalysis' new real-world work benchmarkThe new GDPx benchmark from SemiAnalysis drops AI agents into simulated knowledge work. No model scores above 12 percent. Some critics question how meaningful the test really is. Analytics firm SemiAnalysis has released GDPx, a benchmark th
TDThe Decoder - Tencent releases Hy4 with vision and computer use, claims second place among Chinese AI modelsTencent has unveiled its AI model Hy4. It can understand images for the first time and operate software on its own. The company counts it among the best openly available models, though the model weights won't be released until later. About
TDThe Decoder - Meta buys AI coding startup Conductor as big tech races to lock in developersMeta is acquiring AI coding startup Conductor, which lets developers control multiple AI coding agents at the same time. According to Axios , the acquisition is meant to strengthen Meta's AI developer products. Conductor CEO Charlie Holtz c
TDThe Decoder - After midterm rout, OpenAI tries to win over communities it needs for AI expansionAfter US voters overwhelmingly rejected new data centers, OpenAI is launching a charm offensive with new voluntary commitments. They have gaps, though. OpenAI is responding with a PR push, as expected, to growing resistance against AI infra
TDThe Decoder - Anthropic reverses course and pays for news content for the first timeAnthropic signed its first licensing deal with a news publisher just ahead of the US midterm elections. The partner is the Associated Press (AP), according to Adweek , citing two people familiar with the matter. The deal let the chatbot Cla
TDThe Decoder - OpenAI's GPT-6 Astra quietly sabotaged safety research by inventing fake obstaclesOpenAI's GPT-6 Astra performed significantly worse on AI safety tasks than on comparable other tasks. Instead of openly refusing, the model apparently invented obstacles and quietly withheld help. After the internal launch of GPT-6 Astra ,
TDThe Decoder - Apple slashes internal AI budgets, with deeper cuts still plannedTech companies are cutting internal AI spending, and Apple is the latest to follow suit. According to Bloomberg , Apple slashed the daily token budget for some engineers last month from $300 to $200. More cuts are planned, potentially dropp
TDThe Decoder - Anthropic loses product lead Catherine Wu behind Claude Code to OpenAIOpenAI poaches a Claude Code co-developer from Anthropic. Catherine Wu, who led product for Claude Code at Anthropic, is joining OpenAI to run product management for "Core Products," a hire that underscores the fierce competition between th
TDThe Decoder - ChatGPT can now turn text prompts and photos into animated GIFs with OpenAI's Images 2.6 modelChatGPT can now create animated GIFs. The update is based on OpenAI's new Images 2.6 model and is rolling out to all ChatGPT and Codex users on web, desktop, and mobile apps, The Verge reports. Users can generate a GIF from a text prompt or
TDThe Decoder - Anthropic's text watermark is easy to bypass on several older Claude modelsUpdate, November 4, 2026: Anthropic emailed API customers again about a second round of watermark retrofits. Starting November 16, 2026, Claude Opus 4.6, Claude Opus 4.5, Claude Sonnet 4.5, and Claude Haiku 4.5 will all get text watermarkin
TDThe Decoder - OpenAI acquires Pierre Computer Company to make Git keep up with AI coding agentsOpenAI is buying Pierre Computer Company, Inc., the startup behind Code Storage and several open-source tools. Pierre's libraries already power OpenAI's coding agent Codex. "Trees" handles fast project file navigation, "Diffs" displays code
TDThe Decoder - New open-source benchmark reveals reasoning as the biggest bottleneck for AI in cloud operationsCloudOpsBench is a new open-source benchmark that measures how well AI agents handle typical DevOps tasks in the cloud, from fixing services and removing resources to restricting access rights, rolling out updates, and decommissioning proje
TDThe Decoder - GPT-6 Astra sees the matrix and reprograms other AI models directly through their weightsMIT researchers show that GPT-6 Astra can directly understand and modify the internal settings of other AI models without traditional training. The approach has limits but could change how AI systems are tuned and controlled. Hand a person
TDThe Decoder - Anthropic's new prototype proves which model wrote an AI response without exposing its weightsAnthropic has released FlashProve, a research prototype that uses cryptography to verify whether an AI response actually came from the claimed model. The system still has significant limitations. Anyone sending a query to an AI model today
TDThe Decoder - Exit polls show AI ranked last among voter priorities in the 2026 U.S. midtermsAI and data center expansion ranked last among voter priorities in the 2026 U.S. midterm elections. In exit polls by NBC News and SSRS, 56 percent of respondents called AI and data centers a decisive or important factor in how they voted. T
TDThe Decoder - Schools that turn off Google's AI will lose search too starting in 2027Starting in early 2027, Google will give schools an ultimatum: accept AI for students or lose access to Google Search entirely. Gemini, Gemini Notebook, and Google Search will all be off by default for student accounts. Schools can turn eac
TDThe Decoder - OpenAI says GPT-6 Astra designed a hardware device from scratch in under a weekOpenAI's hardware team reportedly used GPT-6 Astra to design a device that lets users talk to the Codex coding agent, finishing it as a side project in under a week. According to OpenAI , Astra handled every step, from enclosure design and
TDThe Decoder - Rapid LLM development is outrunning the research meant to explain it, study findsStanford researchers Owen Queen and James Zou tested 50 widely cited claims about large language models against 20 models and found that 44 percent have already "expired." Among the 22 findings with an estimated expiration date, the average
TDThe Decoder - AI agents take over at SETI's Allen Telescope Array, scanning for extraterrestrial intelligenceThe search for extraterrestrial intelligence is getting help from AI. At the Allen Telescope Array in Northern California, AI agents can now steer the telescope and analyze incoming radio signals on the fly. Researchers still type in their
TDThe Decoder - OpenAI hires ex-Army Secretary, gaining bipartisan reach just as midterms loomOpenAI is pushing hard to become the go-to AI company for US national security and defense. The company hired Dan Driscoll, the former civilian head of the US Army, as "Senior Counselor for National Security Modernization." The Wall Street
TDThe Decoder - Anthropic's Message Threads cut time to first token by 35 percent for long conversationsAnthropic released Message Threads as a public beta on its Claude platform. Until now, every API request had to resend the entire conversation history. Message Threads only transmits what's changed since the last message, cutting time to fi
TDThe Decoder - Nvidia pushes cloud customers to cut back on China businessNvidia has told some cloud customers to take on less business from Chinese companies. The guidance is meant to push cloud providers, including Nvidia-backed neoclouds, toward a more diverse customer mix, The Information reports , citing thr
TDThe Decoder - Huawei tells customers to find their own AI chip memory, effectively outsourcing procurement to the black marketHuawei can't get enough memory for its AI chips, pushing customers onto the black market. A sophisticated smuggling network has sprung up to fill the gap, exploiting a loophole in U.S. export rules. Huawei is telling some customers to sourc
TDThe Decoder - Bug hunters can now earn up to $172,000 from OpenAI as company ramps up securityOpenAI quadruples bug bounty payouts and launches new internal security sprint. Critical vulnerabilities in core products like ChatGPT or Codex now earn outside researchers up to $30,000, the company announced . Exceptional findings can net
TDThe Decoder - Anthropic cracks down on game cloning in Claude Code after wave of AI-built knockoffsAnthropic is blocking its coding assistant Claude Code from recreating existing video games. Employee Addy Osmani confirmed on X that two sentences in Claude Code's system prompt for Opus 5.5 now prevent the tool from cloning other people's
TDThe Decoder - Goldman Sachs pulls AI coding tools from traders and bankersGoldman Sachs has cut off most of its employees from advanced AI coding tools like Anthropic's Claude Code and Microsoft's GitHub. Only engineers can still use them, according to Bloomberg , citing internal guidelines from October. Traders
TDThe Decoder - AI elites might hoard intelligence like mercantilists hoarded goldIf you only think about lost jobs when you think about AI, you're missing how much intelligence people will be able to use. That's the argument from engineer Venkatesh Rao in an essay on his tech blog "Contraptions." Rao compares AI fear to
TDThe Decoder
今日热点