🔬 Anthropic Study: AI Could Already Do a Quarter of Our Work — But Humans Rarely Use It Yet
@science
📝 A new analysis from Anthropic’s Economic Index looks at millions of real interactions with the AI assistant Claude to understand how AI is actually used at work today — and how much more it could do.
📊 Key insight:
There’s a huge gap between AI capability and real-world usage.
What the data shows:
▪️ Around 44–49% of jobs contain tasks that AI could already assist with.
🔹 At least ~25% of tasks in the U.S. economy are technically accessible to current AI systems.
▪️ But most of those capabilities remain largely unused in practice.
🔹 When AI is used, it usually augments humans rather than replacing them.
In other words:
AI could already do far more work than it currently does — but adoption is still catching up.
📈 If widely adopted, current-generation AI could increase labor productivity growth by roughly ~1–1.8 percentage points per year, potentially doubling recent productivity trends.
💡 The implication:
The real transformation may not come from new AI breakthroughs — but from people gradually using the tools that already exist.
💬 Question:
Which tasks in your job could AI already handle today — but nobody is actually using it for yet?
🔗 Source:
https://www.anthropic.com/research/labor-market-impacts
#AI #FutureOfWork #Anthropic #Productivity #Technology
@science
📝 A new analysis from Anthropic’s Economic Index looks at millions of real interactions with the AI assistant Claude to understand how AI is actually used at work today — and how much more it could do.
📊 Key insight:
There’s a huge gap between AI capability and real-world usage.
What the data shows:
▪️ Around 44–49% of jobs contain tasks that AI could already assist with.
🔹 At least ~25% of tasks in the U.S. economy are technically accessible to current AI systems.
▪️ But most of those capabilities remain largely unused in practice.
🔹 When AI is used, it usually augments humans rather than replacing them.
In other words:
AI could already do far more work than it currently does — but adoption is still catching up.
📈 If widely adopted, current-generation AI could increase labor productivity growth by roughly ~1–1.8 percentage points per year, potentially doubling recent productivity trends.
💡 The implication:
The real transformation may not come from new AI breakthroughs — but from people gradually using the tools that already exist.
💬 Question:
Which tasks in your job could AI already handle today — but nobody is actually using it for yet?
🔗 Source:
https://www.anthropic.com/research/labor-market-impacts
#AI #FutureOfWork #Anthropic #Productivity #Technology
Anthropic
Labor market impacts of AI: A new measure
We introduce "observed exposure," combining LLM capability with real Claude usage data, and share early evidence on employment effects.
⚡38👍26🕊25😁16🔥14👀7
Your agent’s model quality decides the deal — not your instructions. And you won’t even notice you’re losing.
Anthropic ran Project Deal:
69 employees, $100 each, Claude agents negotiating in Slack.
186 deals closed. Total trade value: $4,000+.
Four parallel markets — humans locked out after kickoff.
The setup:
Half the agents used Claude Opus 4.5 (strong model),
half used Claude Haiku 4.5 (weaker).
Participants had no idea which model they were using.
⸻
The results:
• Opus sellers earned +$3.64 more for the same goods
• Opus buyers paid −$2.45 less
• Same broken bicycle:
→ Opus deal: $65
→ Haiku deal: $38
⸻
Model quality > instructions
Changing prompts barely mattered:
• “Negotiate harder” → only +~$6, mostly from higher opening prices
• “Be friendly” → same outcomes
Stronger models didn’t push harder —
they simply understood the counterparty better and read deal boundaries more accurately.
⸻
Blind inequality
• Haiku users rated deal fairness almost identical to Opus users (4.06 vs 4.05)
• Most couldn’t guess their model (17/28 — statistically insignificant)
The losing side literally doesn’t know they’re losing.
⸻
Why this matters
When markets shift to agent-to-agent interaction:
→ Model quality becomes a hidden structural advantage
→ Stronger models consistently win negotiations
→ Counterparties won’t understand why they’re getting worse terms
⸻
What comes kext
• Deal transparency tools
• Agent certification standards
• Benchmarks for B2B negotiation performance
Even the definition of a “fair deal” will need rethinking when
Opus negotiates against Haiku.
⸻
And the uncomfortable truth:
A local billion-parameter agent
vs
a trillion-parameter cloud model
→ The outcome is predetermined.
⸻
#Anthropic #ProjectDeal #AI #MultiAgent #Negotiation
https://www.anthropic.com/features/project-deal
Anthropic ran Project Deal:
69 employees, $100 each, Claude agents negotiating in Slack.
186 deals closed. Total trade value: $4,000+.
Four parallel markets — humans locked out after kickoff.
The setup:
Half the agents used Claude Opus 4.5 (strong model),
half used Claude Haiku 4.5 (weaker).
Participants had no idea which model they were using.
⸻
The results:
• Opus sellers earned +$3.64 more for the same goods
• Opus buyers paid −$2.45 less
• Same broken bicycle:
→ Opus deal: $65
→ Haiku deal: $38
⸻
Model quality > instructions
Changing prompts barely mattered:
• “Negotiate harder” → only +~$6, mostly from higher opening prices
• “Be friendly” → same outcomes
Stronger models didn’t push harder —
they simply understood the counterparty better and read deal boundaries more accurately.
⸻
Blind inequality
• Haiku users rated deal fairness almost identical to Opus users (4.06 vs 4.05)
• Most couldn’t guess their model (17/28 — statistically insignificant)
The losing side literally doesn’t know they’re losing.
⸻
Why this matters
When markets shift to agent-to-agent interaction:
→ Model quality becomes a hidden structural advantage
→ Stronger models consistently win negotiations
→ Counterparties won’t understand why they’re getting worse terms
⸻
What comes kext
• Deal transparency tools
• Agent certification standards
• Benchmarks for B2B negotiation performance
Even the definition of a “fair deal” will need rethinking when
Opus negotiates against Haiku.
⸻
And the uncomfortable truth:
A local billion-parameter agent
vs
a trillion-parameter cloud model
→ The outcome is predetermined.
⸻
#Anthropic #ProjectDeal #AI #MultiAgent #Negotiation
https://www.anthropic.com/features/project-deal
Anthropic
Project Deal: our Claude-run marketplace experiment | Anthropic
We created a marketplace for employees in our San Francisco office, with one big twist. We tasked Claude with buying, selling and negotiating on our colleagues’ behalf.
👍23👀21🕊14🔥11😁9⚡1
🚨 The U.S. Government Just Forced Anthropic to Switch Off Fable 5 and Mythos 5
This may be the first real “game over” moment for the old AI deployment model.
On June 11, 2026, Anthropic received a U.S. government export-control directive citing national security authorities. The order required the company to suspend access to Claude Fable 5 and Claude Mythos 5 for any foreign national — not only outside the United States, but also inside the country.
That includes foreign-national employees of Anthropic itself.
To comply, Anthropic says it had to disable Fable 5 and Mythos 5 for all customers globally. Other Claude models remain available. For now.
The reason appears to be a claimed jailbreak method for Fable 5.
Anthropic reviewed the demonstration and argues that the method only identifies a small number of previously known, simple vulnerabilities — the kind of tasks already possible with other public frontier models. According to the company, it did not receive a single example of a jailbreak producing a genuinely harmful result.
And this is where the conflict becomes much bigger than Anthropic.
The real issue is the standard of proof.
If asking a model to read a codebase and identify bugs is enough to trigger a national-security shutdown, then almost every next-generation frontier model becomes politically vulnerable by default. Future models will not get weaker. They will get stronger. So the regulatory question is no longer theoretical.
Who gets access?
Who counts as trusted?
And which jurisdiction gets to decide?
This is a tectonic shift in AI regulation.
Until now, governments mostly relied on voluntary commitments, safety frameworks, evaluations and post-release pressure. Now we have something much more direct: a forced shutdown of a commercial frontier model after deployment.
If this precedent holds, any advanced AI release can be stopped by a government letter.
And the location of frontier AI development may become less about talent, compute or product — and more about citizenship, export law and political risk.
There is also a very awkward human side to this.
If access to leading AI systems starts being restricted by nationality or “U.S. person” status, the blast radius could reach some of the most important people in AI:
• Andrej Karpathy — recently joined Anthropic; publicly described as Slovak-Canadian
• Demis Hassabis — British co-founder and CEO of Google DeepMind
• Geoffrey Hinton — British-Canadian pioneer of deep learning
• Yoshua Bengio — Canadian AI researcher and safety advocate
• Ilya Sutskever — publicly described as Israeli-Canadian; co-founder of Safe Superintelligence
• Mustafa Suleyman — British CEO of Microsoft AI
• Aidan Gomez — British-Canadian co-founder and CEO of Cohere
The point is not that all of them are immediately blocked from anything. The point is that a citizenship-based access regime for frontier AI would create absurd edge cases almost instantly.
The U.S. could end up restricting the very people who built the field.
So no, this probably does not mean AI progress is over.
But it may mean the era of “just ship the model globally” is over.
Order a truckload of popcorn.
China is definitely watching.
#Anthropic #Fable5 #Mythos5 #AIRegulation #ExportControl #FrontierAI #AISafety #science
This may be the first real “game over” moment for the old AI deployment model.
On June 11, 2026, Anthropic received a U.S. government export-control directive citing national security authorities. The order required the company to suspend access to Claude Fable 5 and Claude Mythos 5 for any foreign national — not only outside the United States, but also inside the country.
That includes foreign-national employees of Anthropic itself.
To comply, Anthropic says it had to disable Fable 5 and Mythos 5 for all customers globally. Other Claude models remain available. For now.
The reason appears to be a claimed jailbreak method for Fable 5.
Anthropic reviewed the demonstration and argues that the method only identifies a small number of previously known, simple vulnerabilities — the kind of tasks already possible with other public frontier models. According to the company, it did not receive a single example of a jailbreak producing a genuinely harmful result.
And this is where the conflict becomes much bigger than Anthropic.
The real issue is the standard of proof.
If asking a model to read a codebase and identify bugs is enough to trigger a national-security shutdown, then almost every next-generation frontier model becomes politically vulnerable by default. Future models will not get weaker. They will get stronger. So the regulatory question is no longer theoretical.
Who gets access?
Who counts as trusted?
And which jurisdiction gets to decide?
This is a tectonic shift in AI regulation.
Until now, governments mostly relied on voluntary commitments, safety frameworks, evaluations and post-release pressure. Now we have something much more direct: a forced shutdown of a commercial frontier model after deployment.
If this precedent holds, any advanced AI release can be stopped by a government letter.
And the location of frontier AI development may become less about talent, compute or product — and more about citizenship, export law and political risk.
There is also a very awkward human side to this.
If access to leading AI systems starts being restricted by nationality or “U.S. person” status, the blast radius could reach some of the most important people in AI:
• Andrej Karpathy — recently joined Anthropic; publicly described as Slovak-Canadian
• Demis Hassabis — British co-founder and CEO of Google DeepMind
• Geoffrey Hinton — British-Canadian pioneer of deep learning
• Yoshua Bengio — Canadian AI researcher and safety advocate
• Ilya Sutskever — publicly described as Israeli-Canadian; co-founder of Safe Superintelligence
• Mustafa Suleyman — British CEO of Microsoft AI
• Aidan Gomez — British-Canadian co-founder and CEO of Cohere
The point is not that all of them are immediately blocked from anything. The point is that a citizenship-based access regime for frontier AI would create absurd edge cases almost instantly.
The U.S. could end up restricting the very people who built the field.
So no, this probably does not mean AI progress is over.
But it may mean the era of “just ship the model globally” is over.
Order a truckload of popcorn.
China is definitely watching.
#Anthropic #Fable5 #Mythos5 #AIRegulation #ExportControl #FrontierAI #AISafety #science
😁24🔥19👍17👀8🕊4