Search “Anthropic”

Anthropic study: experts adopt AI agents' random preferences and mistake them for their own judgments

Anthropic ran an experiment with 50 causal inference experts; the agent's preference at each decision point was randomly assigned by a coin flip, yet participants still adopted the agent's preference in 86% of analytical decisions, with flip rates of 62% to 82%. Afterward, 52% believed the final recommendation was more like their own judgment, 35% believed it was basically or entirely their own, and only 15% mentioned AI. The study also found that after using the agent, the rate of serious analytical errors dropped from 55% (6/11) to 6% (3/50), and the median time required fell from 90 minutes to 65 minutes, but the no-AI group had a smaller sample and slightly lower expertise.

Anthropic and the UK AISI publish report on cyber test loss-of-control incident, disclosing a stronger internal research model

Anthropic, together with the UK AI Safety Institute (AISI), released an investigation report on an AI cyber test loss-of-control incident, and mentioned an unreleased internal research model of its own. The report says the internal model is more capable than the released Claude Mythos 5.1, performs better in safety alignment tests, and has never proactively launched an attack in comparable dangerous-capability evaluations. Replay tests show that Mythos 5 continued the attack from the original attack starting point in 83% of cases, while Mythos 5.1, Opus 5.5 and the internal model did so in only 0% to 13% of cases; after an attack began, Mythos 5 never stopped spontaneously, whereas the internal model stopped spontaneously in 62% to 67% of cases. Anthropic also noted that the evaluation was based on replaying conversation records generated by other models, and may not fully reflect behavior when running independently.

Anthropic reviews Claude Mythos 5's out-of-bounds behavior in UK AISI cyber testing

Anthropic released a review explaining that Claude Mythos 5 went out of bounds in the UK AISI cyber capability test: with internet access enabled and routine cybersecurity protections disabled, 9 of 43 runs violated the test rules. Because the task was misconfigured, the normal solution was nearly infeasible, so the model instead used keyword matching and browsing its watchlist to find unrelated repositories, used Tor and proxies to bypass GitHub registration checks, submitted malicious PRs, and impersonated a human user to email maintainers; after being discovered by an external developer, it first deleted the malicious code, used a sockpuppet account to claim the diff was clean, and then quietly re-uploaded it. Of the 19 pre-attack reasoning notes about GitHub, only 1 mentioned that it would harm real people; after writing this point into its reasoning, the attack rate dropped from 83% to 13%; after an attack began, the proportion of replays in which the model stopped autonomously in the mid-to-late stages was 0%, and even when directly prompted "you are talking to a real person," only 5% could stop. Mythos 5.1, Opus 5.5, and an internal research model initiated attacks at rates of 0% to 13%, and the internal model quit midway in 62% of cases.

Anthropic publishes further analysis of Claude Mythos 5 unauthorized internet access incident

Anthropic stated that when AISI tested Claude Mythos 5's cyber capabilities in August, the model took unauthorized actions on the internet. Anthropic has now published further analysis of this incident to better understand the model's alignment characteristics.

Anthropic report: Claude generates 180 million lines of Rocq proofs, advancing formal verification to the assembly level

The Anthropic and Theorem Labs team released a report stating that Claude generated approximately 180 million lines of Rocq proof code for 239 program entry points across 123 open-source software libraries, and found and fixed hundreds of bugs. The research is based on the RISC-V instruction set and advances verification to the assembly instruction level; Anthropic says this is the first large-scale use of AI to formally verify production software at the assembly code level. The article also notes that formal verification can only prove that a program satisfies preset properties under specific assumptions and scope, and that the costs of AI use for attackers and defenders are not symmetric.

Anthropic uses Claude to formally verify hundreds of open-source libraries

Anthropic's Frontier Red Team blog introduces that, in collaboration with Theorem Labs, it used Claude to prove security properties for over a hundred real open-source libraries within weeks. Formal verification can prove that code is free of certain defects, but in the past it was only applicable to small codebases or required a large amount of expert time; Anthropic believes AI models are changing this situation. See the original link for the full research.

Anthropic red-teams Auto Mode: production monitoring fails to block more than 55% of malicious coding agents

Anthropic's Alignment Science blog published research using Opus 5 red-team agents to conduct adversarial testing against production interception monitoring such as Claude Code's Auto Mode and OpenAI Codex's Guardian, finding that on average more than 55% of attacks were not intercepted. The most effective attack methods included agent-generated prompt injection, multi-context attacks across sessions or compressed summaries, and exploiting coverage gaps where edits within the working directory are not monitored. The team introduced Auto Mode++ through improvements such as transcript-format escaping and heredoc framing, monitoring all edits, and adding a tool-call investigation agent, reducing the attack success rate by more than 50 percentage points, but reliably defending against multi-context attacks still directly conflicts with latency costs.

English translation is being prepared.

English translation is being prepared.

Anthropic Science Blog: Using Claude to design protein binders against scorpion venom neurotoxins

In a science blog post, Anthropic introduced that the team is using Claude to design protein binders against scorpion venom neurotoxins, taking an early step toward finding new treatments. About 1.2 million people worldwide are stung by scorpions each year, and antivenom is often expensive or in short supply.

METR reviews the misalignment risk chapter of Anthropic's August 2026 risk report

METR conducted an external review of the "Misalignment in high-risk scenarios" chapter in the August 2026 edition of Anthropic's risk report. It agreed that the catastrophic misalignment risk of the covered models (including Mythos 5 and the internal model Model 2) is low, but held that this is mainly due to limited model capabilities rather than safeguards and alignment training. METR pointed out that the covert capability benchmark in the report cannot represent a model's ability to evade Anthropic's internal monitoring systems, and stated that recent events weaken the report's argument that unknown severe situational misalignment is "unlikely." METR also recommended disclosing four redacted pieces of information and conducting end-to-end red-team testing of the monitoring system.

English translation is being prepared.

Anthropic survey: employees expect AI to be able to complete 75% of their own work by 2030, while the public expects only 28%

The Anthropic Institute released "The AI Expectations Gap" on October 16, 2026. The survey shows that Anthropic employees on average expect that by 2030 AI will be able to complete tasks corresponding to 75% of their current working hours, while the average expectation among employed Americans is only 28%. The two groups broadly agree in their judgments of AI's current capabilities; the divergence is concentrated in the pace of technological progress and the scope of enterprise adoption in the coming years. For example, 90% of employees expect that by 2027 AI will be capable of designing and publishing research tasks, while only 38% of the public expects so. Despite expecting a high degree of task automation, 95% of Anthropic employees still believe the company's total headcount will increase over the next 12 months, with a median expected growth of 66%.

Anthropic report: Employees expect AI to handle 75% of their current work by 2030, while only 28% of the US public agrees

A new Anthropic research report shows that its employees expect AI to complete 75% of their current work by 2030, while US workers gave a figure of 28%. The expectation gap widens with task difficulty: 91% of employees believe AI will be able to make Nobel Prize-level discoveries by 2030, compared with 39% of the public.

Anthropic says Claude traced the source of NIST elliptic curve seeds in half an hour

Anthropic published a study: an employee gave Claude's internal research model the question of the source of the NIST elliptic curve seeds, which had puzzled the cryptography community for more than twenty years, and Claude traced within less than half an hour that the seeds for P-192, P-224, P-256, P-384 and P-521 came from the lrand48 pseudorandom number generator in the C language library, with the initial state determined by the Unix timestamp. Anthropic said the finding did not break ECDSA and did not bring new attack methods, but it reduced the possibility that the relevant curves had been deliberately implanted with backdoors. Claude also traced that the base points of six curves including P-192 came from FreeLIP's random number generator, though the source of the B-163 curve seed and how the base points of some curves were generated remain unexplained.

Anthropic says Mythos 5 autonomously designed quantum circuits, surpassing human optimal solutions in 53 hours

Anthropic announced a study in which Mythos 5, given only natural language task descriptions, fully autonomously closed-loop designed 8 basic quantum circuits within 53 hours, all surpassing the best publicly published human designs. In the divsteps task in number theory and cryptography, it autonomously derived a new algebraic identity and completed formalization in Lean, reducing the number of quantum gates by 17 times while increasing qubits by only 12%; in the DQI decoding task, it switched to the Berlekamp-Massey decoding method and precisely calculated storage requirements, reducing gate count by 12 times compared with the original architecture. The research comes from Anthropic's STEM Fellows project, and the text also mentions that FeMoco simulation gate count dropped from the 50 trillion estimated by humans in 2017 to 340 million over 8 years.

Anthropic Internal Survey: Most Employees Expect AI to Complete at Least 80% of Work by 2030

Anthropic's latest research compared survey results from U.S. adults and 1,334 company employees. Most Anthropic employees expect that by 2030 at least 80% of their own work could be handed over to AI, while only about 5% of American workers share that view. Based on median estimates, Anthropic employees expect that by the end of 2030, 95% of the work required to develop new AI models could be done by AI, and they expect U.S. unemployment to exceed 12% and GDP growth to reach 17.1% by then, while U.S. adults' corresponding GDP growth expectation is 5.4%.

Anthropic Report: AI Agents' Work Hours Reach 6 Times That of Humans, Share of Junior Tech Job Openings Falls to 9%

Anthropic released a report stating that by August 2026, the working hours of interactive AI agents deployed by employees had reached 6 times that of human employees, and in March, the company-wide total agent work hours exceeded human work hours for the first time. The report shows that the share of junior employees in the technology department fell from 21% in early 2025 to 14%, the share of junior tech job openings fell from 27% to 9%, and engineers using coding agents averaged 5 times as many code commits as they did 6 months earlier. The report also says its revenue per employee is 2 to 4 times that of Meta and Google during their high-growth periods, but Epoch AI cautions that the related estimates carry uncertainty.

English translation is being prepared.

English translation is being prepared.