Loading story…
Anthropic released Claude Sonnet 5.5 for coding and knowledge-work use, claiming more than 30% faster responses than Sonnet 5, lower token use and up to 30% lower per-task costs while retaining pricing of $2 per million input tokens and $10 per million output tokens. The company reported major benchmark gains, including a 70.6% Terminal-Bench 4.0 score versus 10.3% for Sonnet 5, while positioning Opus 5.5 as the stronger option for complex, open-ended work requiring sustained judgment. Sonnet 5.5 is available through the Claude Platform, AWS, Google Cloud and Microsoft Azure, with a zero-data-retention option in supported deployments.
Anthropic assessed Sonnet 5.5 as having cybersecurity capabilities comparable to Opus 5.5 and applied a three-stage, Opus-level cyber-safety policy. Routine defensive development work, including bug discovery and remediation, remains permitted, but higher-risk cyber requests may be blocked or routed to the older Sonnet 5; API customers must explicitly enable that fallback. Anthropic cautioned that the fallback model is materially more susceptible to prompt injection, while classifiers may also generate false positives when processing untrusted repository, web, file, memory or connector content. The company is tuning those controls, plans expanded access for verified defenders through its Cyber Verification Program, and added measures intended to prevent model-capability extraction through large-scale fake-account activity.
Track how attackers are adapting to this technology.
1 event from the most recent confirmed update back to the earliest known activity.
Anthropic launched Claude Sonnet 5.5 for coding and office-work tasks via the Claude Platform, AWS, Google Cloud, and Microsoft Azure. The release added three-stage cyber-safety classifiers and fallback to Sonnet 5 for certain higher-risk cyber requests, while retaining the prior Sonnet API list price.
Vulnerabilities, threat actors, malware, products, organizations, and breaches Mallory has linked to this story.
Follow how adversaries are adapting to this technology, and where it touches your stack today.
3 references tracked. Mallory keeps watching after this page renders.
helpnetsecurity.com
Open sourcethenewstack.io
Open sourcethenewstack.io
Open sourceMap indicators from this story to your assets and identify affected systems in minutes.
Every observed campaign, victim, and pivot linked to actors named in this story.
Malware, exploits, and IOCs connected to the activity described here.
YARA, Sigma, and Snort rules deployed to your SIEM as soon as they’re published.
Get matching new stories delivered to your team as they break — not the next morning.
Ask questions about this story and take action on the answers.
Other stories Mallory is tracking that overlap on entities, sources, or topic labels.
Anthropic released **Claude Opus 5.5**, claiming faster generation, lower token consumption, and improved agentic coding performance over Opus 5. The company said the model delivers more than 30% faster output and lower operating costs—reported as roughly 20% lower API pricing and up to 40% lower cost for many workloads—while increasing subscription five-hour usage limits by 20%. Enterprise evaluations by Box, GitHub, Ramp, and Deloitte reportedly found fewer steps and tokens for development tasks, clearer outputs, and improved code-review bug detection. Cybersecurity, biology, and frontier-LLM-development prompts are screened by safety classifiers. Requests flagged as cybersecurity-related are generally routed transparently to the older, less capable **Opus 4.8** model; biology and frontier-model requests are typically routed to Opus 5. Anthropic expects organizations approved through its Cyber Verification Program to receive Opus 5.5 cybersecurity access in coming weeks, but the routing may create capability inconsistencies in multi-turn production agent workflows.
Sep 25, 2026
Anthropic introduced **Opus 5**, a new flagship model aimed at long-running coding and knowledge-work tasks, positioning it as cheaper, more token-efficient, and less restrictive than some competing systems. Reports said the model delivers incremental gains rather than a major capability jump, but performs strongly on software-development benchmarks, in some cases surpassing **Opus 4.8**, **Fable**, and OpenAI’s **GPT-5.6-Sol** on coding-oriented tasks. Anthropic also launched **Automatic Fallbacks** in beta, which redirect prompts flagged by Opus 5 safety classifiers to **Opus 4.8** instead of rejecting them outright, while retaining zero-retention handling and biology-related safeguards carried over from earlier models. Coverage of the release highlighted that Opus 5’s stronger autonomous behavior changes enterprise security requirements even if Anthropic deliberately limited its cutting-edge cybersecurity training. The model can identify vulnerabilities and support defensive code review, but it reportedly trails **Fable** and especially **Mythos 5** in offensive tasks such as vulnerability exploitation. Analysts noted that organizations deploying Opus 5 for agentic workflows will need tighter operational controls, including isolated execution environments, short-lived credentials, improved telemetry, and semantic circuit breakers to contain runaway actions and unexpected token spend.
Jul 27, 2026
Anthropic announced **Claude Code Security**, an embedded capability in *Claude Code* that scans customer codebases for vulnerabilities and suggests patches, initially rolling out to a limited set of enterprise/team customers for testing. The company said the feature was stress-tested via internal red-teaming, Capture-the-Flag exercises, and collaboration with **Pacific Northwest National Laboratory**, and positioned it as a way to reduce reliance on manual security reviews as AI-assisted “vibe coding” increases and attackers also use AI to accelerate weakness discovery. In parallel, Anthropic released **Claude Sonnet 4.6**, emphasizing improved coding performance, stronger “computer use” capabilities, and expanded developer tooling (e.g., adaptive/extended thinking modes, beta context compaction, and API tools for web search/fetch and code execution). Separate commentary highlighted the security risk of **agentic coding assistants** (e.g., *Claude Code*, *Cursor*, *GitHub Copilot*) operating with broad privileges—file access, shell execution, and secret handling—and argued that the emerging **Model Context Protocol (MCP)** ecosystem needs stronger, future-proof identity controls; additional industry guidance promoted **MLSecOps** as a way to integrate security into AI/ML development lifecycles, though it did not report a specific incident or vulnerability.
Jun 29, 2026