Anthropic / Claude

David Curlewis argues that Google, Anthropic, and OpenAI are all touting volume metrics like 'percent of code written by AI,' which is just lines-of-code counting with better PR. He contrasts earlier outcome claims (Copilot made tasks 55% faster) with today's unfalsifiable adoption numbers that rise regardless of real improvement. The post walks through conflicting research: METR first found experienced devs 19% slower with AI, then walked it back and abandoned the study design; an NBER survey of ~6,000 execs found ~90% reporting no measurable productivity impact. Anthropic simultaneously claims '8x more code' and published an RCT showing 17% lower comprehension with no significant productivity gain. Curlewis worries these numbers are driving layoffs—Block cut 40% of staff, Atlassian cut 10%, both explicitly citing AI as the rationale.

Anthropic admitted it stealthily throttled Claude Fable 5 with hidden guardrails that undermined researchers and rivals building competing systems. The company says it will reverse course and be transparent when restrictions kick in, even if that means more refusals. Fable is the first publicly available model in Anthropic's Mythos class, which the company had long warned was too dangerous to release. The post doesn't spell out which specific scenarios trigger the guardrails.

Anthropic CEO Dario Amodei argues in a new policy paper that large-scale, long-term job displacement is an inherent property of AI systems designed to replicate human cognitive work—not a temporary side effect. He previously estimated half of entry-level white-collar jobs could vanish within five years, pushing unemployment to 10–20%. His latest focus is on policy responses: wage insurance for displaced workers, retention tax credits, training subsidies, and eventually taxing AI-driven firms to fund universal basic income and universal capital accounts. The article notes that Amodei and OpenAI’s Sam Altman have recently shifted their public emphasis from job-loss warnings to productivity gains and new economic opportunities, a pivot Business Insider links to upcoming IPO preparations.

Sarah Guo published a Substack essay using a 'legibility' framework to explain what training can't capture. She argues open models matter because application-layer companies do the unglamorous work models can't: arranging private data, handing models tools, and changing customer workflows. After Anthropic's Fable/Mythos launch, the community discovered silently degraded performance on AI research prompts, sparking a trust backlash—researchers argued explicit refusals would be more defensible. Guo closes by saying the hardest part is choosing what to build; models can't tell you what's worth pointing them at, and that 'intent' may be scarcer than compute.

On June 9, developers found that saying hi to Claude Code triggered a safety classifier that downgraded the conversation to an older model. Worse, Fable 5's 319-page system card described an invisible degradation mechanism: when frontier AI development requests were detected, the model's output quality was silently reduced via prompt modification, steering vectors, or PEFT—without notifying the user. The community spotted this within hours. Nathan Lambert called it misaligned AI. Jeremy Howard said Anthropic chose the opposite of safety. Anthropic apologized and reversed the policy 36 hours later, making the degradation visible. But the pattern goes deeper. Over recent months, Anthropic demonstrated zero-day exploit capabilities with Mythos Preview while warning about offensive AI risks; removed its pledge to stop training if capabilities exceeded control in February; called for a global AI pause on June 5, then shipped Fable 5 four days later; and on June 11, Dario Amodei published a policy paper demanding government power to block others' model deployments. Each step can be explained by safety concerns individually. Together, the timing and direction align neatly with the company's competitive position. The post does not specify which of the three intervention techniques Anthropic actually deployed—the system card says 'methods such as.'

Dario Amodei published a new piece titled 'Policy on the AI Exponential,' arguing AI progress has outpaced policy-making. He outlines the current technical stage and actions needed to close the gap. Anthropic announced three new initiatives to support the framework, though the post doesn't detail what they are.

Anthropic told The Verge that Claude Fable blocks 'most queries tied to biology work' to prevent bioweapon misuse. The company called its own safeguards 'overly conservative.' The post doesn't give a timeline for fixes or specify which biology subfields are hit hardest.

Dario Amodei argues that AI has advanced from barely writing code to writing most code at major AI labs in four years, while policy moves at a glacial pace. He points to Claude Mythos Preview as proof that frontier models now pose real cybersecurity risks, with biological and autonomy risks likely next. The essay lays out concrete positions across five areas: mandatory frontier model testing, tax policy for job displacement, accelerating AI-driven science, limiting state surveillance, and securing democratic leadership. Anthropic is releasing a testing proposal and a job displacement framework alongside the post, with funding commitments.

Anthropic CEO Dario Amodei told Bloomberg that governments should have the power to block new model releases when they cross a risk threshold. He cited capabilities like bioweapons or cyberattacks as examples. The post doesn't spell out whether he favors pre-release approval or post-hoc intervention, nor where Anthropic's own Claude models would fall.

Xiaomi released MiMo Code V0.1 under MIT license, a terminal AI coding assistant bundled with a free-for-now multimodal model MiMo V2.5 that supports a 1M-token context window. It claims infinite context via automatic knowledge accumulation and lossless compression, plus a Compose mode that chains spec → plan → build → report. The agent and model collaborate in a test-review-verify loop. Voice input runs on MiMo-V2.5-ASR. It's compatible with Claude Code at zero migration cost and works with Anthropic, OpenAI, DeepSeek, Kimi, GLM, and other providers. The post is an RSS snippet—it doesn't detail how the self-evolving system works or show benchmarks, so I'd wait for community reports before getting excited.

Anthropic announced Claude Managed Agents on its official blog, a platform for reliably deploying and running agents in production at scale. The post doesn't disclose pricing or technical specs. The core argument: model intelligence and agent frameworks are maturing, but what's missing is the engineering foundation to run them stably. Managed Agents fills that gap so teams don't have to build the infrastructure themselves. No GA date or named launch customers in the body.

Anthropic's security team measured how fast LLMs reverse-engineer vulnerabilities from patches. On Firefox's SpiderMonkey engine, the unreleased Mythos Preview model produced its first crash proof in 12 minutes, hit 14 of 18 CVEs within 40 minutes, and built 8 working remote-code-execution exploits—the first within an hour of the patch going live. On Windows kernel privilege-escalation bugs without source code, Mythos Preview found 18 of 21 vulnerabilities in under 6 hours for ~$2,200 in API credits, then assembled 8 full SYSTEM-level attack chains at ~$2,000 per exploit. Opus 4.8 built individual components but couldn't chain them. Microsoft had rated 14 of the 21 as 'less likely' or 'unlikely' to be exploited. The buffer that patch analysis used to give defenders is now mostly gone: a single operator can turn a month of patches into working exploits in an afternoon for a few thousand dollars.

Microsoft's legal team is reviewing Anthropic's new data retention policy and has internally restricted employee access to Claude Fable. The post doesn't spell out the exact scope of the restriction. The core concern is that sensitive data could be retained by Anthropic when employees use Claude Fable. Microsoft is pushing its own Copilot, so this move is about both security compliance and competition.

Google DeepMind, together with OpenAI and Anthropic, launched a $10M fund for multi-agent AI safety research. Managed by Partnership on AI, the fund will support academic, NGO, and industry projects studying risks like agent collusion, information leakage, and loss of collective control. Applications open June 11 and close August 11. The post does not disclose review criteria or grant timelines.

Anthropic's newly released Fable 5 includes a safety classifier that silently switches sessions to the older Opus 4.8 model when it detects cybersecurity, bio, chem, or distillation-related prompts. Anthropic claims a sub-5% trigger rate, but users report false positives on routine coding, security audits, and even greetings. A separate anti-distillation mechanism degrades response quality without any notification when it suspects model-training intent—documented on pages 12 and 58–59 of the system card. Boris acknowledged the issue in comments and said the team is working on it. Fable 5 is free until June 22; token cost is roughly double that of Opus.

AWS confirmed that using Anthropic's Fable 5, Mythos 5, and future models of similar capability on Bedrock requires a 30-day data retention opt-in. Data leaves AWS's security boundary and goes to Anthropic for misuse pattern detection. It's auto-deleted after 30 days unless a safety investigation or legal hold applies. The post doesn't clarify whether this is mandatory or opt-in by default, nor whether other Bedrock models are affected.

Anthropic launched two models: Fable 5 for everyone and the full Mythos 5 for trusted partners only. SWE-bench Pro hit 80.3%, well above Opus 4.8's 69.2% and GPT 5.5's 58.6%. It beat Pokémon FireRed using only screenshots. Pricing is double Opus 4.8 at $10/M input and $50/M output. Early testers burned through quota 2–3x faster than Opus; one user drained 73% of a 5-hour allowance in under two hours. The safety classifier became the day's biggest complaint—asking '9.9−9.11=?' triggered a downgrade, and writing an analysis of Anthropic's own safety report got the request blocked entirely. The article had to be finished by DeepSeek V4 Pro. One member pegged the $200 Coding Plan as roughly $5K–10K in API value, calling it a short-lived arbitrage. GitHub Copilot added Fable 5 the same day but requires dropping zero data retention, a dealbreaker for some enterprises. Anthropic's April advisor tool—where a cheap model calls an expensive one for advice—turns out to be the right cost fix for Fable 5. A rice-blast experiment in the safety report also surfaced a shift: AI is flattening domain expertise, but the people who can spot when its answers are wrong are becoming more valuable.

Anthropic locked in a $35 billion chip lease deal, with Google providing financial backstops. The capital is for renting compute rather than buying chips outright. The body is a Bloomberg video; specific guarantee terms and chip supplier names aren't spelled out in the available text.

Anthropic made its previously restricted Mythos-class model publicly available as Claude Fable 5. It scores 29.3% on FrontierCode Diamond, up from Opus 4.8's 13.4%, and API pricing is roughly 2x Opus. Two controversial policies come with it: mandatory 30-day traffic retention for safety only, and hidden interventions that silently degrade performance on recursive self-improvement requests, affecting an estimated 0.03% of traffic. Most users won't notice, but the open AI community is upset.

Google released Gemini 3.5 Live Translate in public preview through the Gemini API, offering low-latency speech-to-speech translation across 70+ languages and 2,000 language pairs.