Skip to content
Live
#Anthropic
Nov 4Wed
  3. ITHome · AI58

    Anthropic develops FlashProve open-source system: using cryptography to prove AI output source

    Anthropic is developing an open-source system called FlashProve, which uses cryptography to prove that AI output indeed comes from a specified model, similar to email encryption signatures. The system can prevent AI companies from secretly switching to different models, verify third-party statements about model usage, and does not disclose model weights, supporting closed-source models. The code has been open-sourced on GitHub. Current research is still ongoing and has not yet been deployed.

    Same story · 5 outlets

    How the accounts differ:MarkTechPost states the proof volume is 18.78 MB; The Decoder states it is about 19 MB.

  4. ITHome · AI38

    Anthropic engineers warn: The more accurate AI is, the easier humans lose the ability to think independently, and the more dependent they become, the harder it is to stay alert

    Anthropic Claude Code engineer Addy Osmani warns that reducing AI model output errors makes it harder for humans to stay alert, and may lead to losing independent thinking and verification ability due to reliance on AI, forming a 'cognitive surrender'. He proposes three usage levels: 'cognitive offloading', 'postponing judgment', and 'cognitive surrender', suggesting users form preliminary hypotheses before accepting AI results, actively seek rebuttal evidence, and require Agents to list autonomous decisions and hypotheses.

  6. MarkTechPost71

    Anthropic open-sources FlashProve: using zkSNARK to prove responses from committed weights

    Anthropic's cutting-edge red team has released an open-source research prototype FlashProve, which uses zkSNARK to prove that a response is generated by committed model weights without revealing the weight information. On Llama 4 Maverick (400B MoE), a single H100 achieves a proof speed of approximately 110 tokens/s, with a proof size of 18.78 MB, and CPU verification takes 17 seconds.

    Same story · 5 outlets

    How the accounts differ:MarkTechPost states the proof volume is 18.78 MB; The Decoder states it is about 19 MB.

  8. TechCrunch · AI58

    Anthropic is developing a system called FlashProve to verify the identity of AI models

    Anthropic has released an open-source system called FlashProve, which uses cryptography to verify that specific outputs come from designated AI models, preventing model spoofing or replacement; the system requires AI labs to publish model fingerprints for verification without disclosing model weights, and the code is open-source.

    Same story · 5 outlets

    How the accounts differ:MarkTechPost states the proof volume is 18.78 MB; The Decoder states it is about 19 MB.

  12. TechCrunch · AI65

    Anthropic still faces economic challenges with its computing costs.

    Anthropic's latest earnings report showed the proportion of computing costs rose to 58% of revenue, continuing to expand from 48% in the previous quarter, while revenue reached $17 billion but losses simultaneously increased. The company paid $1 billion in computing fees to SpaceX, Nscale, and others to meet demand, with $500 million in short-term commitments significantly higher than long-term contracts, indicating that computing will likely remain the largest operational expense in the future.

Nov 3Tue
  2. The Decoder65

    Anthropic bans game cloning in Claude Code to combat the wave of AI replicas

    Anthropic prevents its coding assistant Claude Code from replicating existing video games, employee Addy Osmani confirms the addition of two new lines in the Opus 5.5 system prompts prohibiting cloning others' games, this action violates the usage policy.

    Same story · 5 outlets
  3. ITHome · AI60

    Anthropic warns: Claude will soon have recursive self-improvement (RSI) capabilities.

    Anthropic warns in the S-1/A prospectus supplement that Claude is about to develop a model with recursive self-improvement (RSI) capabilities, which may soon reach the 'AI-driven explosive technological advancement' defined in the 'Responsible Scaling Policy'. Currently, Claude has not yet crossed the RSI threshold, but the company commits that in the foreseeable future, the R&D process will retain human involvement and consider revising the policy to ensure that the development of cutting-edge models is always under human control.

    Same story · 2 outlets
  4. Hacker News · 300+65

    Anthropic IPO should be blocked by SEC to protect the public

    Anthropic has filed its IPO registration statement, but financial disclosures show commitments of spending close to 900 billion USD, most of which extend to 2030, and revenue growth in 2026 has already stagnated. The company needs to reverse this trend to achieve extraordinary revenue growth to fulfill its obligations, but weak demand, intense competition, and cost pressures will make this extremely difficult. If it goes public, employees, retail investors, and retirees will bear the debt repayment risk of Anthropic.

    Same story · 13 outlets
  5. ITHome · AI65

    Goldman Sachs restricts traders' use of AI tools like Claude Code, causing dissatisfaction.

    Goldman Sachs bans most employees except engineers from using Anthropic Claude Code and GitHub, causing dissatisfaction among traders who had used these tools to speed up their work. Goldman Sachs says the move is based on efficiency and risk assessment, but traders express frustration, as Goldman Sachs had previously praised non-technical employees for using AI tools.

    Same story · 2 outlets
  6. Anthropic Research62

    Anthropic has released the verifiable reasoning system FlashProve, generating proofs at approximately 110 tokens/second on a single H100.

    Anthropic announced the verifiable reasoning prototype FlashProve, using zkSNARK to prove that the response indeed comes from the committed model weights. For the 400B parameter Llama 4 Maverick (4-bit quantization), a single H100 generates proofs at approximately 110 tokens/second, with proofs for each 1,024 token request being about 19 MB, and verification on a 24-core CPU taking about 17 seconds.

    Same story · 5 outlets

    How the accounts differ:MarkTechPost states the proof volume is 18.78 MB; The Decoder states it is about 19 MB.

    Why it matters: Anthropic has released the measured data and Lean formal proof of the verifiable reasoning prototype FlashProve, allowing for a cross-comparison with existing zkML solutions.

  8. Claude Code releases15

    Claude Code v2.1.317 released

    Claude Code v2.1.317 released, new /schedule task connector warning, network call email address check, SDK callback failure block, desktop settings load notification, sub-agent tool call no response prompt.

  9. 36Kr · AI65

    Goldman Sachs restricts traders and bankers from using AI tools such as Claude Code and GitHub.

    Goldman Sachs has banned the use of most AI tools like Claude Code and GitHub for employees except for engineers, following previous approval for bankers. The ban has caused dissatisfaction among some traders, who had used these tools to monitor trading flows, model, and create content.

    Same story · 2 outlets
  12. Hacker News · AI hot posts

    The internal AI usage cost (retail price) at Anthropic has exceeded employee salaries

    The cost of Claude AI, used internally by Anthropic (calculated based on retail price), exceeded the company's monthly employee salary (excluding stock options) in April and surged 250 times since January in September. The company did not actually spend more cash but evaluated internal AI consumption based on the client's charging price, but this trend highlights the dominant role of AI in software development, with over 90% of the code written by Claude, while revenue growth far exceeds employee expansion.

  13. TechCrunch · AI71

    Anthropic warns of nearing RSI capabilities in the revised S-1

    Anthropic, in the revised S-1A risk factors section submitted on Monday, states that its internal estimates show it may soon meet the definition of 'rapid acceleration of progress brought by AI' in its Responsible Scaling Policy, but has not yet designated any existing models as exceeding that threshold.

    Same story · 2 outlets

    Why it matters: Anthropic admits it may soon hit its self-defined RSI threshold in the revised S-1's risk factors, providing public text on how leading labs assess their safety boundaries.

  14. The Verge · AI65

    Anthropic updated Claude Code's system prompts to prevent AI from replicating specific games

    Anthropic updated Claude Code's system prompts to prevent AI from replicating specific games, but users can still use it to create original games. This move aims to address the current issue of rampant decompilation, reassembly, and porting of AI-generated retro games. Anthropic added relevant instructions in the Opus 5.5 version, emphasizing the creation of original games, specific types of games, or games built based on users' own designs and resources, prohibiting direct copying of existing games.

    Same story · 5 outlets
  15. Ars Technica · AI65

    Anthropic introduces Claude Code new rules to restrict game cloning, but it is difficult to stop the AI game replication wave

    Anthropic has added a new rule 'tengu_heron_brook' in Claude Code v2.1.314 version, prohibiting the model from cloning specific existing games. This rule has been applied to the binary file released on October 30. However, this measure is expected to be difficult to slow down the recent trend of AI-assisted game porting and replication, as the Claude Opus 5.5 model also includes similar restriction instructions.

    Same story · 5 outlets
  16. Hacker News · 300+

    Ubuntu has not fixed the serious kernel vulnerability CVE-2026-98432 that has existed since 2016

    Anthropic has developed an exploit that can remotely exploit the CVE-2026-98432 vulnerability to gain server root privileges. The Linux kernel TCP vulnerability has existed since 2016, Debian has fixed it since October 16, but Ubuntu has not released a patch until November 2. All supported Ubuntu versions (including LTS) are listed as vulnerable, and its official website promises to fix critical vulnerabilities in a timely manner, but the actual response is severely delayed.

Nov 2Mon
  1. Import AI42

    Realtime-Bench released: Benchmark tests to evaluate the performance of LLMs on low-latency tasks

    Realtime-Bench benchmark tests evaluated the performance of LLMs on 50 real-time interactive games and web tasks, with GPT-6 Astra achieving a 36% pass rate under real-time clock conditions; this test was developed by institutions such as the University of Texas at Austin to measure the model's real-time state awareness, adaptive reasoning, and time-aware action selection capabilities.

  2. ITHome · AI60

    Hackers Develop New Malware VOLTA, Using AI to Analyze Whether Victims Are Worth Infiltrating

    Cisco Talos Discovers New Windows Malware VOLTA, Integrating AI Models to Analyze Victim Data to Assess Infiltration Value, Uploads Information to Anthropic Claude for Analysis, and Attempts Self-Destruction to Reduce Tracking Risk. This Makes Attacks More Precise,posing Higher Risk of Network Attacks to Large Organizations and Executives.

  3. ITHome · AI65

    Anthropic is rushing towards an IPO under the codename 'Ants', and entomologists have something to say

    Same story · 13 outlets
  4. ITHome · AI65

    Researchers test AI agent clusters in Minecraft: Collaboration is prone to bureaucracy, higher win rate against GPT in battles

    Researchers used 50 GPT-6.1 Sol agents to build a civilization community in Minecraft, but failed to develop due to bureaucratic tendencies; in a wake-up war test with 10 GPT-6.1 Sol against 10 Claude Opus 5.5, GPT agents adopted a more aggressive strategy, completing problem diagnosis and sharing solutions; in seven Mindustry battles, GPT agents won six and lost one, demonstrating the advantage of swarm mode.