Anthropic unveils Claude Opus 4.7 with enhanced cyber filters

Anthropic Strengthens Its Strategy with Claude Opus 4.7

On April 16, 2026, Anthropic launched its latest artificial intelligence model, Claude Opus 4.7, promising notable improvements in financial analysis and multidisciplinary reasoning. This launch stands out for maintaining pricing despite the technological advancements offered.

Performance and Improvements

Claude Opus 4.7 succeeds its predecessor, Opus 4.6, with enhanced performance in complex tasks. Benchmarks reveal significant progress in automated financial analysis and reasoning, while also providing better long-term memory management and high-definition image analysis.

Cybersecurity Initiative: the Cyber Verification Program

In response to growing cybersecurity concerns, Anthropic has established the Cyber Verification Program. This program allows cybersecurity professionals to submit requests for adjustments to safeguards if their legitimate use is blocked. Users must specify whether they are accessing Claude via the API, the Anthropic app, or Microsoft Azure, and provide their organization ID.

However, organizations under Zero Data Retention (ZDR) contracts cannot participate in this program and must contact a sales representative for any requests.

Enhanced Safeguards for Secure Use

Opus 4.7 incorporates stricter cyber filters than its predecessor, aimed at preventing malicious uses while allowing defensive uses, such as threat detection and malware analysis. Anthropic anticipates that professionals like pentesters or incident response teams may encounter obstacles, hence the implementation of the unlocking form.

Although Claude Opus 4.7 is accessible to the general public, the Glasswing project demonstrates that Anthropic reserves its more powerful models, such as Claude Mythos Preview, for a select group of verified partners.