Search “Anthropic”

Anthropic red-teams Auto Mode blocking monitoring

Anthropic Alignment Science blog publishes research red-teaming production block monitoring such as Claude Code's Auto Mode and OpenAI Codex's Guardian. The tests used an Opus 5 red-team agent, and results showed that on average more than 55% of attacks were not blocked. The research points out that the most effective attack methods include agent-generated prompt injection, multi-context attacks across sessions or compressed summaries, and exploitation of a coverage gap where edits within the working directory are not monitored. The team then made improvements such as transcript format escaping and heredoc framing, monitoring all edits, and adding a tool-call investigation agent, launching Auto Mode ++, and the attack success rate dropped by more than 50 percentage points. However, reliably defending against multi-context attacks still directly conflicts with latency costs.

View story:Anthropic red-teams Auto Mode block monitoring

Anthropic teases proactive agent Claude Duet

Anthropic has teased the upcoming launch of a proactive agent called Claude Duet. According to a Testing Catalog report on October 22, 2026, the agent will support multiple connectors, agentic shopping, and secure sign-in: it can complete purchases for businesses via Stripe's digital wallet Link, check out at online stores using Shopify's Shop Pay, and will support logging into websites with credentials saved in 1Password. The report also noted that Anthropic previously announced Claude Duet would launch as a standalone research preview app, built on Claude Opus 5.5, and made available in the U.S. via a waitlist for the Pro and Max plans. There is currently no qualified factual evidence of further development progress.

Anthropic releases new SDK and ant CLI

On October 21, 2026, Anthropic released a new SDK and ant CLI, defaulting to API version 2026-10-21. According to Claude Devs, this is the first official version release since 2023, and the 2023-06-01 version has not been deprecated. This version removes temperature, top_p, and top_k from the Messages API; documents in assistant turns and citations in user turns will return 400; and it replaces some deprecated Admin and Enterprise Analytics API fields. Developers can use the /claude-api skill in Claude Code to complete the update, or modify items one by one according to the migration guide. There are currently no subsequent developments.

Dario Amodei admits Anthropic mishandled three recent initiatives

Dario Amodei admitted that Anthropic could have handled three recent initiatives better, including the federal AI bill it drafted. He called these three initiatives communication missteps, describing their wording as careless or their content as outdated, and said he takes responsibility as CEO. The same report also noted that he simultaneously stood by Anthropic's acceleration path, including its in-house chip plans, compute deals exceeding $500 billion, and the fact that Claude has taken on 26% of its AI development tasks. Currently, this event rests on only this one piece of factual evidence, with no subsequent developments yet.

Anthropic previews proactive agent Claude Duet

Anthropic has teased the upcoming launch of a proactive agent called Claude Duet. According to a Testing Catalog report on October 22, 2026, the agent will support multiple connectors, agentic shopping, and secure sign-in: it can complete purchases for businesses via Stripe's digital wallet Link, check out at online stores using Shopify's Shop Pay, and will support logging into websites with credentials saved in 1Password. The report also noted that Anthropic previously announced Claude Duet would launch as a standalone research preview app, built on Claude Opus 5.5, and made available in the U.S. via a waitlist for the Pro and Max plans. There is currently no qualified factual evidence of further development progress.

View story:Anthropic teases proactive agent Claude Duet

Anthropic releases Claude API version 2026-10-21

Anthropic releases Claude API version 2026-10-21, which is the first official version release since 2023; version 2023-06-01 is not deprecated. This version removes temperature, top_p, and top_k from the Messages API; documents in assistant turns and citations in user turns will return 400; some deprecated Admin and Enterprise Analytics API fields are replaced. New major versions of the Anthropic SDK and ant CLI are released the same day and use 2026-10-21 by default; you can update in Claude Code with the /claude-api skill, or handle items one by one according to the migration guide. There is currently no further evidence-backed development after this version's release.

Dario writes a long post responding to Anthropic safety questions

In October 2026, Anthropic CEO Dario published a long post on X responding to outside doubts about Anthropic's stance on safety. According to Phoenix Technology, he was responding to questions raised by AI safety investigator Nathan Calvin, and for the first time disclosed that the company had already trained a high-end model at the Mythos level as early as February this year; because its cybersecurity defense mechanisms were not yet complete, it was released only to a very small number of partners, and was not opened to the public until the safety guardrails were in place in June, a gap of about four months. He also listed the costs the company had voluntarily borne for safety, including a large number of false positives caused by biological risk filtering, pushing for legislation such as California SB 53, and a plan to contribute 5% of the company's shares to a public interest fund. Dario said this approach amounts to competing toward a higher standard, while Calvin questioned how much more safety can make the company concede in the race. Other reports said the long post was responding to a list of nine questions raised by outside safety researchers, who questioned that Anthropic had departed from its original safety mission and become a commercial machine sprinting toward a $2 trillion IPO.

Dario publishes long article responding to Anthropic safety questions

In October 2026, Anthropic CEO Dario published a long post on X responding to outside doubts about Anthropic's stance on safety. According to Phoenix Technology, he was responding to questions raised by AI safety investigator Nathan Calvin, and for the first time disclosed that the company had already trained a high-end model at the Mythos level as early as February this year; because its cybersecurity defense mechanisms were not yet complete, it was released only to a very small number of partners, and was not opened to the public until the safety guardrails were in place in June, a gap of about four months. He also listed the costs the company had voluntarily borne for safety, including a large number of false positives caused by biological risk filtering, pushing for legislation such as California SB 53, and a plan to contribute 5% of the company's shares to a public interest fund. Dario said this approach amounts to competing toward a higher standard, while Calvin questioned how much more safety can make the company concede in the race. Other reports said the long post was responding to a list of nine questions raised by outside safety researchers, who questioned that Anthropic had departed from its original safety mission and become a commercial machine sprinting toward a $2 trillion IPO.

View story:Dario writes a long post responding to Anthropic safety questions

Anthropic launches Claude Security, which continuously scans and tracks fixes

On October 21, 2026, Anthropic launched a new version of Claude Security that continuously scans repositories as code changes, automatically marks issues as fixed once fixes go live, and connects to existing tools via API and MCP; it is currently in public beta for Claude Enterprise customers. The new version provides a knowledge base and an editable threat model for each project, on which scans base their judgment of which risks are worth flagging, and gives each finding a severity rating and an attack path description. Anthropic says Claude Security has been used by thousands of organizations since its public beta in April and has cumulatively found more than 1 million vulnerabilities; enterprise customers receive $25,000 in free credits per month for two months, applicable to usage through December 31, 2026. Follow-up reporting further explains that after code changes land it automatically rescans, automatically closes the corresponding issue once a patch goes live, reopens it if the vulnerability reappears, and combines security team notes and the editable threat model to determine which vulnerabilities attackers can actually reach; previously it was only an on-demand or scheduled scanner whose results went stale as soon as code changed.

Report says Eric Schmidt holds a 1% stake in Anthropic

According to Bloomberg, former Google CEO Eric Schmidt holds about a 1% stake in Anthropic. TechCrunch AI reported on October 21, 2026 that Schmidt participated in Anthropic's Series A funding round in 2021, investing tens of millions of dollars; this stake is about half of the 2% held by co-founder and CEO Dario Amodei. The report also said Anthropic is widely expected to go public before the end of the year, with a maximum valuation expected to reach $2 trillion, at which point the value of Schmidt's stake could increase by as much as $20 billion; neither Schmidt nor Anthropic immediately responded to requests for comment. The next day, Fortune AI reported, citing people familiar with the matter cited by Bloomberg, that if Anthropic goes public at a target valuation of $2 trillion, this stake could be worth $20 billion; the report said Schmidt was introduced to Dario Amodei by Cami Clark and participated in Anthropic's $124 million Series A funding round in 2021, when the company was valued at $461 million. There have been no further developments so far.

Anthropic releases Claude API version 2026-10-21

Anthropic releases Claude API version 2026-10-21, which is the first official version release since 2023; version 2023-06-01 is not deprecated. This version removes temperature, top_p, and top_k from the Messages API; documents in assistant turns and citations in user turns will return 400; some deprecated Admin and Enterprise Analytics API fields are replaced. New major versions of the Anthropic SDK and ant CLI are released the same day and use 2026-10-21 by default; you can update in Claude Code with the /claude-api skill, or handle items one by one according to the migration guide. There is currently no further evidence-backed development after this version's release.

View story:Anthropic releases Claude API version 2026-10-21

Anthropic releases new SDKs and ant CLI

On October 21, 2026, Anthropic released a new SDK and ant CLI, defaulting to API version 2026-10-21. According to Claude Devs, this is the first official version release since 2023, and the 2023-06-01 version has not been deprecated. This version removes temperature, top_p, and top_k from the Messages API; documents in assistant turns and citations in user turns will return 400; and it replaces some deprecated Admin and Enterprise Analytics API fields. Developers can use the /claude-api skill in Claude Code to complete the update, or modify items one by one according to the migration guide. There are currently no subsequent developments.

View story:Anthropic releases new SDK and ant CLI

Anthropic launches Claude Security that continuously scans and tracks remediation

On October 21, 2026, Anthropic launched a new version of Claude Security that continuously scans repositories as code changes, automatically marks issues as fixed once fixes go live, and connects to existing tools via API and MCP; it is currently in public beta for Claude Enterprise customers. The new version provides a knowledge base and an editable threat model for each project, on which scans base their judgment of which risks are worth flagging, and gives each finding a severity rating and an attack path description. Anthropic says Claude Security has been used by thousands of organizations since its public beta in April and has cumulatively found more than 1 million vulnerabilities; enterprise customers receive $25,000 in free credits per month for two months, applicable to usage through December 31, 2026. Follow-up reporting further explains that after code changes land it automatically rescans, automatically closes the corresponding issue once a patch goes live, reopens it if the vulnerability reappears, and combines security team notes and the editable threat model to determine which vulnerabilities attackers can actually reach; previously it was only an on-demand or scheduled scanner whose results went stale as soon as code changed.

View story:Anthropic launches Claude Security, which continuously scans and tracks fixes

WSJ sees Anthropic's draft IPO filing

The Wall Street Journal has seen Anthropic's draft IPO filing. A report dated October 21, 2026, said the draft showed Anthropic had first-half revenue of $16.3 billion and a book loss of $80.7 billion, with compute commitments of about $450 billion; the draft also disclosed fixed payments totaling $25.2 billion to Google, AWS and Microsoft over the next 7 to 10 years, payable whether or not the capacity is used. A report dated October 22 said the same draft showed first-half revenue of $16.3 billion, up nearly 14-fold year over year, and a net loss of $80.7 billion, of which about $72 billion came from the revaluation of financing instruments such as convertible notes; compute and infrastructure spending in the same period was nearly $9 billion, about 55% of revenue, and forward compute commitments exceeded $600 billion, potentially reaching as high as $750 billion if additional agreements are finalized. There is currently no further progress confirmed by factual evidence.

Dario Amodei admits Anthropic mishandled three recent initiatives

Dario Amodei admitted that Anthropic could have handled three recent initiatives better, including the federal AI bill it drafted. He called these three initiatives communication missteps, describing their wording as careless or their content as outdated, and said he takes responsibility as CEO. The same report also noted that he simultaneously stood by Anthropic's acceleration path, including its in-house chip plans, compute deals exceeding $500 billion, and the fact that Claude has taken on 26% of its AI development tasks. Currently, this event rests on only this one piece of factual evidence, with no subsequent developments yet.

View story:Dario Amodei admits Anthropic mishandled three recent initiatives

WSJ saw Anthropic's draft IPO filing

The Wall Street Journal has seen Anthropic's draft IPO filing. A report dated October 21, 2026, said the draft showed Anthropic had first-half revenue of $16.3 billion and a book loss of $80.7 billion, with compute commitments of about $450 billion; the draft also disclosed fixed payments totaling $25.2 billion to Google, AWS and Microsoft over the next 7 to 10 years, payable whether or not the capacity is used. A report dated October 22 said the same draft showed first-half revenue of $16.3 billion, up nearly 14-fold year over year, and a net loss of $80.7 billion, of which about $72 billion came from the revaluation of financing instruments such as convertible notes; compute and infrastructure spending in the same period was nearly $9 billion, about 55% of revenue, and forward compute commitments exceeded $600 billion, potentially reaching as high as $750 billion if additional agreements are finalized. There is currently no further progress confirmed by factual evidence.

View story:WSJ sees Anthropic's draft IPO filing

English translation is being prepared.

English translation is being prepared.

View story:English translation is being prepared.

Anthropic discloses 211 scorpion venom neurotoxin-binding proteins designed by Claude

On October 20, 2026, Anthropic disclosed 211 binding proteins designed by Claude against scorpion venom neurotoxins, along with their data. According to Anthropic Research, an internal research model autonomously designed minibinders in Claude Science, generating 318,484 designs within 64 hours and selecting 180, plus 31 added by the team, for a total of 211 submitted to Adaptyv Bio and Twist Bioscience for experimental testing, of which 17 met the binder criteria. All 17 targeted a single toxin; none of the multi-toxin designs met the criteria. The strongest candidates, Cn2-B1 and Cn2-B2, bound Cn2 with single-digit nanomolar affinity, and competition experiments showed they bind near the same epitope as the known neutralizing antibody LR. The study has not yet verified neutralizing efficacy in cells or live animals, and Anthropic does not call them antivenom.

Anthropic discloses 211 scorpion venom neurotoxin-binding proteins designed by Claude

On October 20, 2026, Anthropic disclosed 211 binding proteins designed by Claude against scorpion venom neurotoxins, along with their data. According to Anthropic Research, an internal research model autonomously designed minibinders in Claude Science, generating 318,484 designs within 64 hours and selecting 180, plus 31 added by the team, for a total of 211 submitted to Adaptyv Bio and Twist Bioscience for experimental testing, of which 17 met the binder criteria. All 17 targeted a single toxin; none of the multi-toxin designs met the criteria. The strongest candidates, Cn2-B1 and Cn2-B2, bound Cn2 with single-digit nanomolar affinity, and competition experiments showed they bind near the same epitope as the known neutralizing antibody LR. The study has not yet verified neutralizing efficacy in cells or live animals, and Anthropic does not call them antivenom.

View story:Anthropic discloses 211 scorpion venom neurotoxin-binding proteins designed by Claude