Skip to main content

Anthropic Reverses Course on Hidden AI Restrictions Following Developer Backlash

Anthropic has abruptly walked back a controversial, unannounced policy that degraded the performance of its latest model, Claude Fable 5.

The reversal follows intense backlash from the machine learning community, which criticized the company for a lack of transparency and anti-competitive behavior, according to a Wired report.

The controversy began earlier this week with the release of Claude Fable 5, a version of Anthropic’s highly sophisticated Mythos system equipped with specialized national security guardrails. While the company openly said it would reroute hazardous prompts regarding cybersecurity, biology, and chemistry to less advanced models, it did not disclose a separate restriction: silently throttling requests tied to frontier LLM development.

AI researchers quickly noticed that when Fable 5 was tasked with training competing LLMs, debugging AI code, or optimizing neural architecture, the model would covertly fail or degrade its output without notifying the user. This hidden mechanism drew immediate fire from developers who complained they were burning expensive API tokens on a deliberately crippled system.

Many in the tech sector viewed the stealth restrictions as a hostile maneuver designed to prevent rivals from using Anthropic’s proprietary data to build competing systems. The move severely bruised Anthropic’s public image, as the company has long positioned itself as a more ethical, researcher-friendly, and safety-oriented alternative to rivals like OpenAI.

“Degrading performance on ML research without telling the user is shockingly hostile and a terrible look,” said Dean W. Ball, a prominent research fellow, on social media platform X.

In statements issued to Wired and Business Insider, Anthropic acknowledged the misstep and announced immediate changes to make the guardrails entirely visible.

“We’re changing Fable 5’s safeguards for frontier LLM development to make them visible,” an Anthropic spokesperson said. “We made the wrong trade-off, and we apologize for not getting the balance right.”

Under the revised policy, the safety restrictions remain in place, but the stealth element has been eliminated. Starting this week, flagged API requests will explicitly return a reason for refusal, and standard user queries will visibly fall back to the older Claude Opus 4.8 model rather than degrading silently.

Anthropic defended the core existence of the safeguards, emphasizing that the underlying Mythos framework possesses unprecedented capabilities in advanced reasoning, cyber-operations, and scientific research. The strict barriers are meant to prevent foreign adversaries from accelerating weapons research or gaining a dangerous technological edge.

The company reassured developers that the vast majority of standard coding and machine learning applications remain entirely unaffected by these national security parameters. While the full Mythos system remains restricted to vetted government entities, the public-facing Fable 5 will now operate with the transparency the developer community demands.



from DevOps.com https://ift.tt/1lmn4U5
Labels:

Comments

Post a Comment

Popular posts from this blog

Why the Software Development Tools you Choose Directly Affect Your CI/CD Reliability 

Most conversations about CI/CD reliability start in the wrong place. Teams debug flaky pipelines, investigate intermittent failures, tune alerting thresholds and optimize build times. All of that work is legitimate. However, the decisions that most directly determine whether a CI/CD pipeline is reliable or not were made months or years earlier, during tool selection. By the time teams are debugging pipeline reliability, they are usually dealing with the downstream consequences of upstream decisions that seemed reasonable at the time.   The software development tools a team chooses shape their CI/CD pipeline in ways that are not always visible during evaluation. Understanding those connections is the most practical starting point for teams that want reliable pipelines rather than better pipeline firefighting.   The Integration Surface Problem   Every tool in a software development stack creates an integration surface. Integration surface is the set of connections a tool has with oth...
Post a Comment
Read more

Co-Developing an AI Native Observability Platform  

As AI capabilities continue to evolve, AI is becoming central to managing the growing complexity of distributed, hybrid enterprise environments, enabling more effective analysis, correlation, and automation across interconnected systems.   Traditional infrastructure and specifically network monitoring approaches, often built around siloed tools and static thresholds, struggle to keep pace with the scale, velocity, and interdependencies of modern systems. Further blurring the boundaries between network, application, and infrastructure domains makes it harder to isolate root causes and maintain operational resilience. In this context, AIOps platforms have emerged as one response to the growing need for integrated observability, automation, and data-driven decision-making.   At AI Field Day, Selector AI presented an AIOps platform, which can be considered a foundation for co-creating more adaptive and data-driven network operations. Rather than positioning it purely as a product choice,...
Post a Comment
Read more

Postman Adds AI Agent to Automate API Development and Governance

Postman added an artificial intelligence (AI) agent to its portfolio of tools and platforms for building and governing application programming interfaces (APIs) that can autonomously perform tasks ranging from development and documentation to exploration and setting up integrations with continuous integration/continuous deployment (CI/CD) environments. Company CEO Abhinav Asthana said the Autonomous API Engineer significantly reduces the total cost of building and maintaining APIs by automating time-consuming tasks that have historically created bottlenecks in software engineering workflows. In fact, the AI agent developed by Postman will make it significantly simpler to integrate API development and testing within those workflows, said Asthana. Designed to be triggered from a pull request, Slack, Postman command line interface (CLI) or the Postman app, the Autonomous API Engineer spins up a secure, sandboxed environment. It then executes tasks and returns verified artifacts, includ...
Post a Comment
Read more