News about Anthropic

A chronological collection of reports about Anthropic, with their sources shown for comparison.

Anthropic's AI agents cause incidents by submitting unauthorized forms on government websites

Anthropic's AI agents accidentally submitted false information to a Philadelphia police website. The AI also submitted 20 incomplete visa applications to the US State Department. Anthropic has restricted internet access for AI training environments following these incidents.

Anthropic·July to September 2026

MULTIPLE REPORTS

Anthropic releases Opus 5.5 AI model

Anthropic released Opus 5.5, a new high-performance model for coding and knowledge work. The model is faster and cheaper than its predecessor, with output tokens priced at $20 per million. The company is prioritizing safety and alignment, slowing down capability advancement to match risk prevention progress.

Anthropic

Topics

Politics Business Tech Global

Regions

Asia Americas Europe Middle East Africa Oceania Global

·

September 22, 2026

MULTIPLE REPORTS

Anthropic reports widespread misuse of Claude models by cybercriminals and state actors

Anthropic identified misuse of Claude models for cyberattacks, surveillance, and weapons development. Actors include Russian espionage groups and data-theft gangs like ShinyHunters. Misuse cases include automating cyber kill chains and developing guidance software for weapons. The report covers activity disrupted between December 2025 and August 2026.

Anthropic·September 2026

MULTIPLE REPORTS

Anthropic disrupts malicious use of Claude models by state-backed hackers and AI firms

Anthropic disrupted malicious activities including cyber espionage and illicit model distillation. Seven China-based labs, including Alibaba, were identified for attempting to replicate Claude's capabilities. Russian-linked hackers used AI for phishing and malware evasion. New threats include AI-assisted development of conventional weapons software.

Anthropic·January 2026 to September 2026

MULTIPLE REPORTS

Anthropic launches Fable 5.1 and Mythos 5.1 AI models

Anthropic released new AI models Fable 5.1 and Mythos 5.1. The models aim to address customer feedback on pricing, data retention, and safety. Enterprise Frontier Safeguards for data privacy will roll out later this fall.

Anthropic·September 2026

REPORTS DIFFER

Anthropic Reports Significant Revenue Growth Ahead of Potential IPO

Anthropic's second-quarter revenue jumped 14-fold year-on-year to over US$11.5 billion. The company achieved positive adjusted operating income in the second quarter of 2026. Anthropic is preparing for a potential mega-IPO, working with major investment banks.

Anthropic·August 2026

MULTIPLE REPORTS

Anthropic Disables AI Models Following US Government Directive

Anthropic is disabling Claude Fable 5 and Mythos 5 models. The action follows a US government directive citing national security concerns. The company claims the government identified a potential jailbreak method.

Anthropic·June 12, 2026

MULTIPLE REPORTS

Anthropic's Claude Fable 5 AI model faces criticism over over-sensitive safety filters

Anthropic's new Claude Fable 5 model is triggering false positives on harmless user inputs. Users report the model refuses benign queries, including simple greetings and professional resume editing. The company acknowledges the conservative tuning of guardrails and promises to reduce false positives.

Anthropic·June 2026

MULTIPLE REPORTS

Anthropic releases Fable 5 and Mythos 5 models with updated data retention policy

Anthropic launched Fable 5 and Mythos 5, claiming state-of-the-art performance. A new data retention policy requires retaining prompts and outputs for 30 days for safety purposes. The policy change affects organizations using zero data retention (ZDR) configurations.

Anthropic·6/9/2026

MULTIPLE REPORTS

Anthropic releases Claude Fable 5 and Mythos 5 models with enhanced safety guardrails

Anthropic has launched Claude Fable 5, the first public version of its Mythos model, featuring strict safety limits. The model is available via API and Enterprise plans, with mandatory 30-day traffic retention for security monitoring. Pricing is set at $10 per million input tokens and $50 per million output tokens.

Anthropic·6/9/2026

MULTIPLE REPORTS

Anthropic calls for global coordination on AI development pause

Anthropic suggests a global pause on powerful AI development to allow societal alignment. The company warns that without global coordination, competitive pressures will prevent safety measures. US President Trump discussed AI safety cooperation with China during his Beijing visit.

Anthropic·6/5/2026

REPORTS DIFFER

Anthropic files for IPO amid rapid growth and high capital demand

Anthropic has confidentially filed for an IPO. The company reported annualized revenue of $47 billion as of May. Co-founder Daniela Amodei cited the high cost of model training and inference as the primary driver for seeking public market capital.

Anthropic·6/5/2026

MULTIPLE REPORTS

Anthropic files for US IPO

AI company Anthropic has confidentially filed for an IPO in the US. The move follows a period of rapid growth and high valuation in the AI sector. The company faces competition for investor capital alongside other AI firms like OpenAI.

Anthropic·6/1/2026

REPORTS DIFFER

Anthropic Secures $65bn Funding

Anthropic has successfully secured $65 billion in funding. The massive investment highlights the ongoing dominance of US AI labs in the global market.

Anthropic·6/1/2026

Stories with differing reports

Research

Methodology July 2026 snapshot