Price:

AI & TECH

Fable 5 triggers shift from AI tasks to autonomous engineering responsibilities

Thursday, July 16, 2026 · from 4 podcasts
AI Daily BriefPeter McCormack ShowThe EconomistDwarkesh Podcast
  • Claude Fable 5 outperforms prior models by wide margins, handling multi-hour engineering loops with minimal human oversight.
  • Anthropic aggressively nerfs biology and AI research queries, angering developers and open-source researchers.
  • AI’s white-collar disruption will hit graduates, rebalancing societal status toward skilled trades.

Anthropic’s Fable 5 model is a benchmark-topping tool for coding, but the real shift is conceptual: developers are moving from giving tasks to delegating responsibilities. According to The AI Daily Brief, Fable 5 more than doubled the performance of its predecessor on Frontier Code, a benchmark that filters for code quality that can actually be merged into production. This isn’t about writing a single function. Early testers, like Stripe, reported that Fable 5 compressed a two-month, multi-engineer migration of a 50-million-line Ruby codebase into a single day. Anthropic’s Felix Ryberg argues this signals a third era - from asking questions, to assigning tasks, to giving autonomous responsibilities like monitoring crash reports.

“Users are moving from giving tasks to giving responsibilities.”

- Felix Ryberg, The AI Daily Brief

This leap in capability coincides with a new economic model. On June 23rd, Fable 5 will be removed from Anthropic’s flat-rate Claude Pro subscription and shift to usage-based pricing, forcing power users to become ‘token efficiency optimizers.’ But the model’s power is constrained by its guardrails. The AI Daily Brief notes Fable 5 triggers automatic fallbacks for basic biology terms like ‘mitochondria’ and is explicitly nerfed for frontier AI research, including tasks like building pre-training pipelines. Researchers like Nathan Lambert and Will Brown see this as a ladder being pulled up, prioritizing proprietary protection over open research needs.

“The model is now effectively off-limits for legitimate open-source research. It marks a shift where labs prioritize protecting their intellectual property over the research community's needs.”

- Will Brown, The AI Daily Brief

This technological pivot carries broad socioeconomic weight. On The Peter McCormack Show, analyst David Goodhart argued this wave of AI will do to graduate office workers what globalization did to factory labor. He contends the West has overproduced people with generalist, clerical skills that AI can now replicate, targeting junior legal and accountancy roles. The displacement, he suggests, could rebalance societal status toward skilled, physical trades that AI lacks the dextery for, like welding or plumbing.

Goodhart’s analysis ties the technological shift to a deeper cultural and political rift - the divide between mobile, liberal ‘Anywheres’ and rooted, traditional ‘Somewheres.’ For years, he argues, policy has favored the graduate class while neglecting the skills and stability valued by the rooted majority. AI’s erosion of the graduate elite’s economic fortress may, ironically, build empathy for communities hollowed out by earlier waves of automation.

This story has not seen significant new development in the days since these analyses. The consensus holds that the frontier of AI engineering is moving from execution to autonomy, but its economic and safety frameworks are creating new friction and social realignments.

#AI Infrastructure#Enterprise#Labor
Nathan LambertPeter McCormackAnthropicStripeClaude

Source Intelligence

- Deep dive into what was said in the episodes

The AI Daily Brief: Artificial Intelligence News and Analysis
The AI Daily Brief: Artificial Intelligence News and Analysis

Nathaniel Whittemore

5 AI Engineering Trends for Non-Engineers • Jul 15

  • Anthropic launched Claude Fable 5, its first Mythos-class model, positioning it above Opus. Mythos 5 lacks Fable’s controversial safeguards and is initially only available to Project Glasswing partners.
  • Nathaniel Whittemore calls Fable 5 the best AI model ever available but notes exploiting state-of-the-art models now requires more than simple prompts.
  • Fable 5 dominates benchmarks: 78% on ExploitBench versus GPT55’s 34%, 66% on HealthBench versus GPT55’s 51.8%, 13.3% on the legal agent benchmark versus GPT55’s 2.1%, and 1932 on GDP Val’s knowledge work test versus Opus 48’s 1890.
  • The model excels at agentic coding: 80.3% on Swebench Pro versus GPT55’s 58.6%, 88% on Terminal Bench versus GPT55’s 83.4%, and 29.3% on Frontier Code versus Opus 48’s 13.4%. Fable scored 91% on Every’s Senior Engineer benchmark.
  • Artificial Analysis found Fable 5 topped its blended benchmark run, overtaking Opus 48 and GPT55, though some noted the overall gap was only five points.
  • Fable 5 scored 72.9% on Cursor Bench, eight points above the previous best, but is more expensive on that cost-performance test.
  • Cognition’s Frontier Code benchmark aims to assess real-world coding quality for merging into production, not just passing unit tests. Meter found more than half of Swebench results are unmergeable 'slop'.
  • Users report Fable flags basic biology terms like 'mitochondria' and 'cancer' as biosecurity risks, switching to Opus 48. Creo, Daria Anupmas, and Fernando documented these blocks.
  • Anthropic’s system card reveals new interventions limiting Fable’s effectiveness for frontier LLM development tasks like building pre-training pipelines or ML accelerator design, aiming to prevent aiding competitors.
  • Researchers Ellie Bau, Nathan Lambert, Dean Ball, and Will Brown criticize the invisible nerfing of AI research capabilities, calling it sad, misaligned, hostile, and a barrier to open model research.
  • Anthropic mandates a 30-day data retention and review policy for Mythos-class models on all platforms. Mike Taylor warns this violates NDAs if memory is on, pulling historical chats into review.
  • Users debate token efficiency: Theo and Chubby hit usage limits quickly, while Tyler Willis, Alex Vulkoff, and Fabio Jonathan argue Fable is not crazily token-hungry and can be cheaper due to one-shot solutions.
  • Ali K Miller says Fable 5 transformed her weekends, calling it an 'actual leap' that autonomously solved a tricky MBA-level word math problem with zero babysitting.
  • Riley Brown one-shot a Swift app replicating Replit mobile with four prompts, prompting debates about AGI claims versus the infrastructure work behind real companies.
  • Stripe reported Fable 5 compressed months of engineering into days, performing a codebase-wide migration on a 50-million-line Ruby project in a day versus a team’s two months.
  • Todd Saunders described Fable building a fully working product feature in real-time during a customer call, creating an 'autonomous looped building' workflow.
  • Felix Ryberg argues Fable 5 initiates a third AI era: moving from asking questions to assigning tasks, and now to giving responsibilities like autonomous loops monitoring crash reports.
  • Nate B. Jones argues the critical new skill is 'task imagination' - conceiving projects that leverage models capable of running for days, which most users currently lack.
Also from this episode: (4)

Models (3)

  • API pricing for Fable 5 is $10 per million input tokens and $50 per million output tokens, double Opus’s cost but lower than some expected. Mythos preview within Project Glasswing costs more than double.
  • Anthropic will remove Fable from subscription plans on June 23rd, moving to usage-based pricing. Whittemore calls this evidence of a 'firmly usage-based pricing paradigm'.
  • Whittemore notes Fable 5 can push back and disagree strategically, then update its position without fully collapsing, making AI-backed ideation more valuable.

Safety (1)

  • Fable 5 has strict guardrails, automatically routing queries about cybersecurity, biology, chemistry, or distillation to Opus 48. Anthropic says 95% of sessions don’t trigger a fallback but is 'hardcore' about biology/chemistry filters.
#Models#Coding#Agents#Enterprise#AI Infrastructure
The Peter McCormack Show
The Peter McCormack Show

Peter McCormack

#193 - David Goodhart - AI Is Coming For The Graduate Elite • Jul 14

  • David Goodhart argues Britain has overproduced people with generalist academic skills. AI will dismantle this class, targeting middling white-collar jobs like junior clerical legal and accountancy work.
  • David Goodhart proposes AI may redistribute status toward skilled trades. He cites welders earning £60,000-70,000 within three years of training.
  • Britain creates tech unicorns, ranking third globally behind China and the US. Andy Haldane cites this, but many firms sell to American buyers due to market size and capital constraints.
  • UK pension funds invest just 4% in domestic equities. Jeremy Hunt and Rachel Reeves both introduced reforms to encourage investment, yet the figure remains unchanged.
Also from this episode: (12)

Education (1)

  • Tony Blair’s 1999 policy pushed 50% of school leavers to university, creating a liberal graduate elite. Goodhart says this left non-graduates feeling like second-class citizens.

Macro (1)

  • Britain's economy has stagnated for 15 years, with few state-delivered improvements. Peter McCormack questions whether any single thing has actually gotten better.

Society (1)

  • Goodhart's 'Anywhere vs. Somewhere' framework splits society. Anythings (25-30% of population) are mobile, liberal graduates; Somewheres (40-50%) are rooted, experience social change as loss.

Politics (9)

  • According to Goodhart, Anywhere overdominance created populist blowback. This imbalance explains Brexit, Trump, Reform UK polling around 30-35%.
  • Goodhart lists failures of British anywhere governance: education favoring degrees over vocational skills, indifferent regional policy, disastrous family policy, and uncontrolled immigration.
  • Britain's manufacturing is 8% of the economy; exports to the EU account for just 2% of national income. Goodhart argues Brexit’s economic impact was inherently limited.
  • McCormack believes the UK economy is designed to 'take' rather than generate wealth. He argues excessive regulation and high marginal tax rates crush small businesses.
  • Goodhart states the UK has the most progressive tax system in Europe. The top 10% face higher tax burdens relative to the middle 10% than even Scandinavian countries.
  • Goodhart claims Blair-Cameron overshot liberalism, dispersing power from Westminster to courts and regulators. This created a 'deep state' blob that paralyzes action.
  • McCormack argues internet expression is under threat. He cites proposals for YouTube to prioritize mainstream news and the potential for internet ID verification.
  • Goodhart advocates a 10-15 year immigration pause to allow integration. He notes Britain’s Muslim population is 6%, but warns its political divergence is worrying.
  • Peter McCormack identifies ethnonationalism as a downstream risk of unchecked multiculturalism. He cites Lebanon’s collapse but says the UK lacks stomach for mass repatriation.
#Models#Labor#Macro#Regulation#Society
The Intelligence from The Economist
The Intelligence from The Economist

The case of the missing totem: Aung San Suu Kyi • Jul 14

  • Turkey is experiencing an economic influx due to the Iran war, with cargo volumes through its biggest port tripling since the Strait of Hormuz closed. Overland logistics and pipelines are also busier.
  • Turkey passed a major tax bill in April 2025 exempting wealthy foreigners from tax on foreign income, a move specifically aimed at attracting Gulf investors. Officials are aggressively pitching Istanbul as a new financial hub.
  • Kerryanne Richmond-Jones notes Turkey's economy has improved from a 'macroeconomic disaster,' with growth at 3.6% and inflation falling by 24 percentage points in 2025, though it remained high at 35%.
  • Despite the influx, Kerryanne Richmond-Jones argues Istanbul is far from becoming the next Dubai or Riyadh due to lack of government investment, investor distrust of Erdogan's economic policies, and competition from established global hubs.
Also from this episode: (10)

Politics (6)

  • Aung San Suu Kyi has been imprisoned by Myanmar's military junta since the 2021 coup, with her last confirmed sighting on December 30, 2022. The junta denies independent verification of her condition despite foreign pressure.
  • Aung San Suu Kyi's original 33-year prison sentence was recently reduced to 27 years. She previously spent about 15 years under detention between 1990 and 2010.
  • Aaron Connolly notes Aung San Suu Kyi's deeply flawed legacy in office, where she defended the military against genocide charges for atrocities against the Rohingya, jailed journalists, and isolated Myanmar, pushing it closer to China.
  • The Myanmar junta refuses to provide proof of life for Aung San Suu Kyi despite requests from her son and foreign leaders like India's Narendra Modi and the UN envoy. Junta leader Min Aung Hlaing reacts with anger to questions about her.
  • Aaron Connolly argues the junta fears Aung San Suu Kyi's talismanic power over the Burmese people more than armed resistance. He cites an example where a party member was jailed for offering alms to monks in her name on her 81st birthday.
  • Releasing Aung San Suu Kyi would likely melt resistance to normalizing relations with the junta at the UN and ASEAN, where Myanmar is suspended from high-level meetings and represented by a civilian government ambassador.

Health (2)

  • Sleep research shows a U-shaped relationship between health and sleep duration, where both too little and too much sleep increase health risks. Major 2015 studies recommended 7-9 hours for adults under 65 and around 7 hours as optimal.
  • A May 2025 Nature study using UK Biobank data and biological clocks found the optimal sleep range is 6.5-7.8 hours for women and 6.4-7.7 hours for men. The U-shaped pattern affected nine organs including the brain, lungs, liver, and skin.

Mental Health (1)

  • Psychiatrist Michael Grandner cautions that sleep duration is just one factor and cause-effect is hard to disentangle. He advises against fixating on exact numbers, noting that worrying about sleep is a major cause of insomnia.

War (1)

  • The IMF estimates the Iran war has caused about $600 billion in damage within Iran itself. The conflict may reduce GDP growth in the broader Middle East by two percentage points.
#War#Diplomacy#Trade#Health#Macro
Dwarkesh Podcast
Dwarkesh Podcast

Adam Brown – A deep but accessible introduction to general relativity • Jul 10

  • Adam Brown argues the most important unanswered question in science is how the human brain achieves high sample efficiency and general capabilities with far less data than modern LLMs. His meta-level take is that neuroscience needs a technological power-up to answer it.
  • Brown's personal hunch is that AI has neglected complex, developmentally staged loss functions. Evolution encodes a specific learning curriculum through many different loss functions, which could be the key to the brain's efficiency.
  • Brown suggests the cortex might be an omnidirectional inference engine, predicting any subset of variables from any other subset, unlike LLMs which are natively optimized only for next-token prediction.
  • Brown outlines Steve Byrnes' theory that the brain's learning subsystem learns to predict the innate responses of a separate steering subsystem, wiring abstract concepts like 'spider' to primitive reflexes like flinching and enabling generalization.
  • Brown notes the human genome is only about 3 GB, a small fraction of which codes for the brain. This compactness is plausible if evolution mainly writes 'Python code' for specific reward functions and bootstrapping rules, not the entire learned model.
  • Brown says current LLM training uses a 'really dumb' form of reinforcement learning without value functions, which is surprising it works so well. In contrast, parts of the basal ganglia may implement simple model-free RL, while the cortex builds a model-based system.
  • A key disadvantage of biological brains is they cannot be copied or externally read, unlike digital models. Advantages include energy efficiency, collocation of memory and compute, and hardware co-designed for potential stochastic, sampling-based inference.
  • Brown states that creating a competent, misaligned agent like a 'paperclip maximizer' likely requires only minimal innate drives for curiosity and exploration, not the full suite of human social instincts. This is an alignment concern.
  • Brown advocates for massively scaling up neuroscience to get a 'ground truth,' specifically by driving down the cost of connectomics. The Welcome Trust estimated the first mouse brain connectome would cost billions; E11 Bio aims to reduce it to tens of millions.
  • Brown describes a moonshot idea of 'behavior cloning' or brain-regularized AI, where models are trained not just on labels but also to predict internal brain activity patterns. This could shape representations and improve generalization, but requires scalable brain scanning tech.
  • On automated theorem proving, Brown says RL from formal verification, as in Lean, will automate the mechanical parts of math. The harder challenge is automating the conceptual creativity of conjecturing interesting new theorems, which might require a loss function for explanatory power.
Also from this episode: (1)

Science (1)

  • Single-cell atlas data shows many more diverse and bespoke cell types in subcortical steering regions like the hypothalamus than in the cortex. Brown interprets this as evidence that evolution's genomic complexity is spent wiring innate reward functions, not the general learning algorithm.
#Models#Reasoning#Brain#Biology

Related Stories