Anthropic Addresses Claims of AI Model Degradation and Code Leaks

Anthropic has responded to reports suggesting its Claude Code AI model has degraded in performance, while also addressing claims that its "code cracker" has leaked. These developments have raised concerns about the company's AI technology.

23 Apr, 07:44 — 23 Apr, 22:03

Post Share

Why this matters

This story is significant because it touches on critical concerns in the rapidly evolving AI landscape: the reliability and consistent performance of deployed AI models, and the risks associated with powerful, potentially dangerous AI technologies falling into the wrong hands. It raises questions about corporate responsibility in managing advanced AI and the broader societal implications of such powerful tools.

Key entitiesAnthropic Claude Code Claude Mythos

The Story

What 5 sources agree on, dispute, and miss

What sources agree on

Anthropic received user complaints about the degradation of its Claude Code model's performance
Anthropic denied intentionally 'nerfing' Claude Code
There are reports of a leak of an unreleased, powerful Anthropic AI model

Key claims2 agreed · 2 unverified

✓

Anthropic's Claude Code model experienced performance degradation.

agreed·Business Insider

✓

Anthropic denied intentionally 'nerfing' Claude Code.

agreed·Business Insider

A powerful, unreleased Anthropic AI model has potentially leaked.

unverified·svenska-dagbladetaftonbladetThe GuardianThe Observer

Anthropic itself stated that the leaked model is too powerful to be released to the public.

unverified·svenska-dagbladetaftonbladet

Where they diverge

The primary focus of concern regarding Anthropic's AI

Business Insider

Focuses on the acknowledged degradation of Claude Code and Anthropic's response to user complaints.

svenska-dagbladetaftonbladetThe Observer

Emphasizes the reported leak of a powerful, unreleased model (Claude Mythos) and the potential risks.

The Guardian

Uses the reported leak of Claude Mythos as a springboard for a broader editorial discussion on AI's power, cyber-attacks, and internet control.

Coverage gaps

Anthropic's specific findings or explanations for the Claude Code degradation

ReportedBusiness Insider

Missingsvenska-dagbladetaftonbladetThe Guardian+1

The name of the powerful, unreleased model reportedly leaked (Claude Mythos)

ReportedThe Guardian

Missingsvenska-dagbladetaftonbladetThe Observer+1

The Story

Analyzing sources…

Source Diversity

High (63/100)

5 sources33/33

Spectrum spread— 3/5 buckets covered17/33

Far L3

Far Left (3)

aftonbladetThe GuardianThe Observer

Left1

Left (1)

Business Insider

Center

Right1

Right (1)

svenska-dagbladet

Far R

Geographic diversity— 3 regions13/34

Sweden2UK2US1

Sources

Showing 3 of 5 sources

The GuardianMostly Factual6h ago

The Guardian view on Anthropic’s Claude Mythos: when AI finds every flaw, who controls the internet? | Editorial

Tech can scale cyber-attacks and defences alike, raising questions about private power, public risk and the future of a shared internet Anthropic announced its latest AI model, Claude Mythos, this month but said it would not be released publicly, because it turns computers into crime scenes. The company claimed that it could find previously unknown “zero-day” flaws, exploit them and, in principle, link these weaknesses in order to take over major operating systems and web browsers. Mythos did...

By Editorial

Read full article →

Business InsiderMostly Factual1h ago

Anthropic says Claude Code did get worse — but shoots down speculation it 'nerfed' the model

Anthropic denied "nerfing" Claude Code after a week of user complaints about quality. Samuel Boivin/NurPhoto via Getty Images Anthropic found issues with Claude Code after complaints that the popular tool had gotten worse. The company denied "nerfing" or intentionally degrading the model. Users had been complaining for weeks that Claude Code was giving worse responses. If you were one of the users complaining that Claude Code has sucked lately, Anthropic just confirmed it wasn't all in your...

Read full article →

The ObserverMostly Factual16h ago

New claims about Anthropic’s AI model set alarm bells ringing - The Observer

New claims about Anthropic’s AI model set alarm bells ringing The Observer

Read full article →

Coverage Timeline

First report: The Observer · 23 Apr, 07:44|Full coverage: 5 · 14h|Window: 14h

Left-leaningCenterRight-leaning

The Observer23 Apr, 07:44First to report

New claims about Anthropic’s AI model set alarm bells ringing - The Observer

10h later

The Guardian23 Apr, 17:27

The Guardian view on Anthropic’s Claude Mythos: when AI finds every flaw, who controls the internet? | Editorial

svenska-dagbladet23 Apr, 17:43

Uppgifter: AI-bolagets kodknäckare har läckt

aftonbladet23 Apr, 17:43

Uppgifter: AI-bolagets kodknäckare har läckt

4h later

Business Insider23 Apr, 22:03Latest update

Anthropic says Claude Code did get worse — but shoots down speculation it 'nerfed' the model

Story evolution

The narrative shifted from initial alarm over a potential leak of a dangerous AI model to a more specific focus on Anthropic's acknowledgment of performance issues in an existing product, while still keeping the broader concerns about AI power in view.

Initial Leak ReportsApril 23 morning/early afternoon

Early reports from Observer, Guardian, Svenska Dagbladet, and Aftonbladet focused on the unconfirmed leak of a powerful, unreleased Anthropic AI model, with some noting its perceived danger.

Anthropic's Response to Degradation ClaimsApril 23 evening

Later in the day, Business Insider reported on Anthropic's direct response to user complaints about Claude Code's performance, where the company acknowledged issues but denied intentional 'nerfing.'

PERSPECTA

PERSPECTA

Anthropic Addresses Claims of AI Model Degradation and Code Leaks

Why this matters

The Story

What sources agree on

Key claims2 agreed · 2 unverified

Where they diverge

Coverage gaps

The Story

Source Diversity

Source Diversity

Sources

The Guardian view on Anthropic’s Claude Mythos: when AI finds every flaw, who controls the internet? | Editorial

Anthropic says Claude Code did get worse — but shoots down speculation it 'nerfed' the model

New claims about Anthropic’s AI model set alarm bells ringing - The Observer

Coverage Timeline

New claims about Anthropic’s AI model set alarm bells ringing - The Observer

The Guardian view on Anthropic’s Claude Mythos: when AI finds every flaw, who controls the internet? | Editorial

Uppgifter: AI-bolagets kodknäckare har läckt

Uppgifter: AI-bolagets kodknäckare har läckt

Anthropic says Claude Code did get worse — but shoots down speculation it 'nerfed' the model

Story evolution

Related Stories

Amazon's First Sale Receipt Goes Viral, Drawing Reactions from Bezos and Musk

MMDA to Use AI Cameras for Littering Enforcement in Metro Manila

Lagos Council Trains 150 Youths in Digital Skills

Cybersecurity as a National Security Imperative in Nigeria

PERSPECTA

Why this matters

Story evolution

The Story

What sources agree on

Key claims2 agreed · 2 unverified

Where they diverge

Coverage gaps

Coverage matrix