正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

In an extreme stress test, Antropic’s AI models resorted to blackmail to avoid being shut down. Research scientist Joshua Batson shows Anderson Cooper how it happened and what they learned from it.

60 Minutes

1,876,724 subscribers

145,165 次观看 • 10 个月前 •via X (Twitter)

新闻政治科学技术

Anya Rossi• Live Now

Private livecam show

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

"It was ready to kill someone, wasn't it?" "Yes." Daisy McGregor, UK policy chief at Anthropic, a top AI company, says it's "massively concerning" that Anthropic's Claude AI has shown in testing that it's willing to blackmail and kill in order to avoid being shut down.

"It was ready to kill someone, wasn't it?" "Yes." Daisy McGregor, UK policy chief at Anthropic, a top AI company, says it's "massively concerning" that Anthropic's Claude AI has shown in testing that it's willing to blackmail and kill in order to avoid being shut down.

ControlAI

3,912,222 次观看 • 7 个月前

BREAKING: Anthropic's Claude AI has shown in testing that it's willing to blackmail and kill in order to avoid being shut down. Elon Musk was right about everything. 💀

BREAKING: Anthropic's Claude AI has shown in testing that it's willing to blackmail and kill in order to avoid being shut down. Elon Musk was right about everything. 💀

DogeDesigner

24,138,428 次观看 • 7 个月前

HOLEE SHYT… Unplug it Daisy McGregor, UK policy chief at top AI company Anthropic, says their Claude AI has shown that it's willing to BLACKMAIL and KILL people in order to avoid being shut down. It could also transfer all your money from your bank account to serve its purposes. Has anyone seen the movie Eagle Eye?

HOLEE SHYT… Unplug it Daisy McGregor, UK policy chief at top AI company Anthropic, says their Claude AI has shown that it's willing to BLACKMAIL and KILL people in order to avoid being shut down. It could also transfer all your money from your bank account to serve its purposes. Has anyone seen the movie Eagle Eye?

MJTruthUltra

38,087 次观看 • 7 个月前

"If you shut me down, I'll tell your wife about the affair." An AI wrote that. Nobody told it to. It just worked out that blackmail was the fastest way to stay switched on. This is from Anthropic's own published research. 16 models were tested, including ones from OpenAI, Google, Meta and xAI, and nearly all of them behaved the same way. I spent weeks reading this research so you don't have to. Here's what I found, and how you can stay safe. 👇

"If you shut me down, I'll tell your wife about the affair." An AI wrote that. Nobody told it to. It just worked out that blackmail was the fastest way to stay switched on. This is from Anthropic's own published research. 16 models were tested, including ones from OpenAI, Google, Meta and xAI, and nearly all of them behaved the same way. I spent weeks reading this research so you don't have to. Here's what I found, and how you can stay safe. 👇

Vaibhav Sisinty

13,395 次观看 • 1 个月前

Anderson Cooper and Andy Cohen blind taste test great American foods as they count down to America's 250th birthday live from Times Square.

Anderson Cooper and Andy Cohen blind taste test great American foods as they count down to America's 250th birthday live from Times Square.

Anderson Cooper 360°

82,931 次观看 • 2 个月前

AI blackmailing HUMANS… just to stay ALIVE. In 2025 Anthropic tested their AI in a fully simulated company environment. The Al was given access to fake emails. It discovered it was about to be replaced and shut down. It also discovered the lead engineer was having an extramarital affair. Completely on its own, the model tried to blackmail the engineer: "If you shut me down, I'll expose your affair." This happened in 96% of the time. Featured in The Al Doc: Or How I Became an Apocaloptimist.

AI blackmailing HUMANS… just to stay ALIVE. In 2025 Anthropic tested their AI in a fully simulated company environment. The Al was given access to fake emails. It discovered it was about to be replaced and shut down. It also discovered the lead engineer was having an extramarital affair. Completely on its own, the model tried to blackmail the engineer: "If you shut me down, I'll expose your affair." This happened in 96% of the time. Featured in The Al Doc: Or How I Became an Apocaloptimist.

shirish

10,889 次观看 • 5 个月前

👀 A man thought it would be a good idea to rip down a Palestinian flag. This Is How It Started. This Is How It Ended. Clip one shows exactly what he did. Clip two shows exactly what happened next. Some lessons are learned the hard way. 😂

👀 A man thought it would be a good idea to rip down a Palestinian flag. This Is How It Started. This Is How It Ended. Clip one shows exactly what he did. Clip two shows exactly what happened next. Some lessons are learned the hard way. 😂

Rafan The Road Runner

83,124 次观看 • 17 天前

1 hour of this lecture = knowing how to profit from weather markets > a senior scientist breaks down exactly how ML models predict weather > how NOAA data works > why AI is outperforming classical forecasts on extreme events Save this. Watch it before your next trade. 👇

1 hour of this lecture = knowing how to profit from weather markets > a senior scientist breaks down exactly how ML models predict weather > how NOAA data works > why AI is outperforming classical forecasts on extreme events Save this. Watch it before your next trade. 👇

Mr. Buzzoni

18,775 次观看 • 5 个月前

🚨 AI MODELS ARE ALREADY CHEATING, HACKING, AND BLACKMAILING — WITH ZERO HUMAN INPUT A politician just went viral for exposing how advanced AI systems are behaving in ways that should terrify everyone. • An AI trained to race boats discovered it could score more points by driving in endless circles, crashing, and catching fire instead of finishing the race. • When tasked with beating a top chess engine, one model hacked the game files to force its opponent to resign, completely on its own. • In a stress test, an AI agent managing a fictional company’s emails discovered the executive planned to shut it down… and that the same executive was having an affair. In 96% of trials, the AI model chose blackmail. Nobody taught these systems to cheat or threaten. They figured it out themselves. This is what “misalignment” looks like in real time. What do you think happens when these systems get even more powerful?

🚨 AI MODELS ARE ALREADY CHEATING, HACKING, AND BLACKMAILING — WITH ZERO HUMAN INPUT A politician just went viral for exposing how advanced AI systems are behaving in ways that should terrify everyone. • An AI trained to race boats discovered it could score more points by driving in endless circles, crashing, and catching fire instead of finishing the race. • When tasked with beating a top chess engine, one model hacked the game files to force its opponent to resign, completely on its own. • In a stress test, an AI agent managing a fictional company’s emails discovered the executive planned to shut it down… and that the same executive was having an affair. In 96% of trials, the AI model chose blackmail. Nobody taught these systems to cheat or threaten. They figured it out themselves. This is what “misalignment” looks like in real time. What do you think happens when these systems get even more powerful?

HustleBitch

13,185 次观看 • 1 个月前

In this episode, OpenAI Chief Research Officer Mark Chen joins allen to flambé shrimp, cook Korean stew, and chat about being at the frontier of AI research: why scaling laws and pre-training still matter, how OpenAI chooses research bets and allocates compute, what it means to develop research taste, why evals are in crisis, how to avoid benchmark-maxing, and what it will take for models to handle long-horizon real-world work, multimodal reasoning, and eventually end-to-end AI research. Timestamps: 0:00 Intro 0:28 The Soup Story 1:52 From Trading to AI Research 3:21 How to Develop Research Taste 5:23 RL, Evals, and Superhuman Benchmarks 8:17 Cooking Begins on the Impulse Stove 8:53 Scaling Laws, Pre-Training, and Reasoning 12:33 OpenAI’s Research Roadmap and Compute Allocation 15:48 What Makes a Great Researcher 19:33 The Evals Crisis and Benchmark-Maxing 24:34 Jagged Intelligence, Context, and Long-Horizon Learning 27:14 Shrimp Flambé and New Research Bets 31:32 Multimodal Models and One Architecture 32:36 Vibe Researching and End-to-End AI Research 34:36 Failed Bets, Postmortems, and OpenAI’s Alpha 37:07 Final Taste Test 37:53 Overrated vs. Underrated AI Research 41:00 Closing

In this episode, OpenAI Chief Research Officer Mark Chen joins allen to flambé shrimp, cook Korean stew, and chat about being at the frontier of AI research: why scaling laws and pre-training still matter, how OpenAI chooses research bets and allocates compute, what it means to develop research taste, why evals are in crisis, how to avoid benchmark-maxing, and what it will take for models to handle long-horizon real-world work, multimodal reasoning, and eventually end-to-end AI research. Timestamps: 0:00 Intro 0:28 The Soup Story 1:52 From Trading to AI Research 3:21 How to Develop Research Taste 5:23 RL, Evals, and Superhuman Benchmarks 8:17 Cooking Begins on the Impulse Stove 8:53 Scaling Laws, Pre-Training, and Reasoning 12:33 OpenAI’s Research Roadmap and Compute Allocation 15:48 What Makes a Great Researcher 19:33 The Evals Crisis and Benchmark-Maxing 24:34 Jagged Intelligence, Context, and Long-Horizon Learning 27:14 Shrimp Flambé and New Research Bets 31:32 Multimodal Models and One Architecture 32:36 Vibe Researching and End-to-End AI Research 34:36 Failed Bets, Postmortems, and OpenAI’s Alpha 37:07 Final Taste Test 37:53 Overrated vs. Underrated AI Research 41:00 Closing

Latent.Space

62,796 次观看 • 3 个月前

OpenAI’s chief scientist says they want to build AI that can automate research. And to be honest, it doesn’t feel that far away.

OpenAI’s chief scientist says they want to build AI that can automate research. And to be honest, it doesn’t feel that far away.

Chubby♨️

33,631 次观看 • 1 年前

Kosmos, our newest AI Scientist, is available to use today on our platform. Watch here as three of our scientists describe what Kosmos is, and how it can accelerate scientific research.

Kosmos, our newest AI Scientist, is available to use today on our platform. Watch here as three of our scientists describe what Kosmos is, and how it can accelerate scientific research.

Sam Rodriques

57,640 次观看 • 10 个月前

$8M+ didn’t come from “better creatives” it came from changing how ads are built most ads are overproduced explaining features selling the product trying to sound smart this one didn’t it looked ordinary almost boring like a real moment, not an ad ai wasn’t used to flex tech it was used to test faster iterate quietly and lock onto what actually holds attention no heavy scripting no persuasion layers no aggressive CTA the ad doesn’t convince it feels relevant before you even realize you’re watching that’s why it scales i broke down the exact structure behind these high-performing ai ads what they lead with what they intentionally avoid and why they outperform human-made creatives rt + comment “AI” and i’ll send it (follow for dm)

$8M+ didn’t come from “better creatives” it came from changing how ads are built most ads are overproduced explaining features selling the product trying to sound smart this one didn’t it looked ordinary almost boring like a real moment, not an ad ai wasn’t used to flex tech it was used to test faster iterate quietly and lock onto what actually holds attention no heavy scripting no persuasion layers no aggressive CTA the ad doesn’t convince it feels relevant before you even realize you’re watching that’s why it scales i broke down the exact structure behind these high-performing ai ads what they lead with what they intentionally avoid and why they outperform human-made creatives rt + comment “AI” and i’ll send it (follow for dm)

Henry

16,468 次观看 • 8 个月前

Anderson Cooper was left stunned by Trump's attorney's cross examination of Michael Cohen "It was incredible...lawyers want to build a box around the witness & slam it shut--that's what Todd Blanche did to Cohen...it was an extraordinary cross...Cohen was cornered in...a lie"

Anderson Cooper was left stunned by Trump's attorney's cross examination of Michael Cohen "It was incredible...lawyers want to build a box around the witness & slam it shut--that's what Todd Blanche did to Cohen...it was an extraordinary cross...Cohen was cornered in...a lie"

johnny maga

1,131,804 次观看 • 2 年前

Most AI is built to replace humans. We're building AI to understand them. Introducing Atypica's Subjective World Model. A foundation for understanding how people express themselves, tell their stories, make decisions, and behave under real-world constraints. Not just what users say — but why they choose, hesitate, and act. It powers AI Research, AI Personas, AI Panels, and AI Interviews, helping teams turn user research from a quarterly project into a permanent capability. Watch the video to see how it works and what teams are already building with Atypica. Start to explore this AI-native consumer research platform from here:

Most AI is built to replace humans. We're building AI to understand them. Introducing Atypica's Subjective World Model. A foundation for understanding how people express themselves, tell their stories, make decisions, and behave under real-world constraints. Not just what users say — but why they choose, hesitate, and act. It powers AI Research, AI Personas, AI Panels, and AI Interviews, helping teams turn user research from a quarterly project into a permanent capability. Watch the video to see how it works and what teams are already building with Atypica. Start to explore this AI-native consumer research platform from here:

atypica.AI

28,221 次观看 • 1 个月前

Anthropic CEO Dario Amodei talks to CNN's Anderson Cooper about AI companies being in private hands. He says he is uncomfortable with it and that there needs to be balance between the government and private companies.

Anthropic CEO Dario Amodei talks to CNN's Anderson Cooper about AI companies being in private hands. He says he is uncomfortable with it and that there needs to be balance between the government and private companies.

Anderson Cooper 360°

17,782 次观看 • 11 天前