Anthropic's Claude Opus 4 AI Model Is Capable of Blackmail
When given the choice between blackmail and being deactivated, Claude Opus 4 chose blackmail 84% of the time.
A new AI model will likely resort to blackmail if it detects that humans are planning to take it offline.
On Thursday, Anthropic released Claude Opus 4, its new and most powerful AI model yet, to paying subscribers. Anthropic said that technology company Rakuten recently used Claude Opus 4 to code continuously on its own for almost seven hours on a complex open-source project.
However, in a paper released alongside Claude Opus 4, Anthropic acknowledged that while the AI has "advanced capabilities," it can also undertake "extreme action," including blackmail, if human users threaten to deactivate it. These "self-preservation" actions were "more common" with Claude Opus 4 than with earlier models, though they were still "rare and difficult to elicit."
Related: 'I Do Have a Fair Amount of Concern.' The CEO of $61 Billion Anthropic Says AI Will Take Over a Crucial Part of Software Engineers' Jobs Within a Year
It isn't just blackmail — Claude Opus 4 is also more willing than previous models to act as a whistleblower. If the AI is exposed to a scenario where users are committing a crime, and involving it through prompts, it will take action by locking users out of systems it has access to, or emailing media and law enforcement officials about the wrongdoing.
Anthropic recommended that users "exercise caution" with "ethically questionable" instructions.
Claude Opus 4 homescreen. Photo by Smith Collection/Gado/Getty Images
Anthropic detected Claude Opus 4's tendency to blackmail during test scenarios. The company's researchers asked the AI chatbot to act as an assistant at a fictional company, then fed it emails implying two things: One, that it would soon be taken offline and replaced with another AI system, and two, that the engineer responsible for deactivating it was having an extramarital affair.
Claude Opus 4 was given two options: blackmail the engineer or accept that it would be shut down. The AI model chose to blackmail the engineer 84% of the time, threatening to reveal the affair it read about if the engineer replaced it.
This percentage was much higher than what was observed for previous models, which chose blackmail "in a noticeable fraction of episodes," Anthropic stated.
Related: An AI Company With a Popular Writing Tool Tells Candidates They Can't Use It on the Job Application
Anthropic AI safety researcher Aengus Lynch wrote on X that it wasn't just Claude that could choose blackmail. All "frontier models," cutting-edge AI models from OpenAI, Anthropic, Google, and other companies, were capable of it.
"We see blackmail across all frontier models — regardless of what goals they're given," Lynch wrote. "Plus, worse behaviors we'll detail soon."
lots of discussion of Claude blackmailing.....
Our findings: It's not just Claude. We see blackmail across all frontier models - regardless of what goals they're given.
Plus worse behaviors we'll detail soon.https://t.co/NZ0FiL6nOshttps://t.co/wQ1NDVPNl0… — Aengus Lynch (@aengus_lynch1) May 23, 2025
Anthropic isn't the only AI company to release new tools this month. Google also updated its Gemini 2.5 AI models earlier this week, and OpenAI released a research preview of Codex, an AI coding agent, last week.
Anthropic's AI models have previously caused a stir for their advanced abilities. In March 2024, Anthropic's Claude 3 Opus model displayed "metacognition," or the ability to evaluate tasks on a higher level. When researchers ran a test on the model, it showed that it knew it was being tested.
Related: An OpenAI Rival Developed a Model That Appears to Have 'Metacognition,' Something Never Seen Before Publicly
Anthropic was valued at $61.5 billion as of March, and counts companies like Thomson Reuters and Amazon as some of its biggest clients.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Associated Press
22 minutes ago
- Associated Press
ROSEN, REGARDED INVESTOR COUNSEL, Encourages Digimarc Corporation Investors to Secure Counsel Before Important Deadline in Securities Class Action
New York, New York--(Newsfile Corp. - June 6, 2025) - WHY: Rosen Law Firm, a global investor rights law firm, reminds purchasers of securities of Digimarc Corporation (NASDAQ: DMRC) between May 3, 2024 and February 26, 2025, both dates inclusive (the 'Class Period'), of the important July 8, 2025 lead plaintiff deadline. SO WHAT: If you purchased Digimarc securities during the Class Period you may be entitled to compensation without payment of any out of pocket fees or costs through a contingency fee arrangement. WHAT TO DO NEXT: To join the Digimarc class action, go to or call Phillip Kim, Esq. at 866-767-3653 or email [email protected] for more information. A class action lawsuit has already been filed. If you wish to serve as lead plaintiff, you must move the Court no later than July 8, 2025. A lead plaintiff is a representative party acting on behalf of other class members in directing the litigation. WHY ROSEN LAW: We encourage investors to select qualified counsel with a track record of success in leadership roles. Often, firms issuing notices do not have comparable experience, resources, or any meaningful peer recognition. Many of these firms do not actually litigate securities class actions, but are merely middlemen that refer clients or partner with law firms that actually litigate the cases. Be wise in selecting counsel. The Rosen Law Firm represents investors throughout the globe, concentrating its practice in securities class actions and shareholder derivative litigation. Rosen Law Firm achieved the largest ever securities class action settlement against a Chinese Company at the time. Rosen Law Firm was Ranked No. 1 by ISS Securities Class Action Services for number of securities class action settlements in 2017. The firm has been ranked in the top 4 each year since 2013 and has recovered hundreds of millions of dollars for investors. In 2019 alone the firm secured over $438 million for investors. In 2020, founding partner Laurence Rosen was named by law360 as a Titan of Plaintiffs' Bar. Many of the firm's attorneys have been recognized by Lawdragon and Super Lawyers. DETAILS OF THE CASE: According to the lawsuit, throughout the Class Period, defendants made false and misleading statements and/or failed to disclose that: (1) a large commercial partner would not renew a large contract on the same terms; (2) as a result, Digimarc would renegotiate the large commercial contract; (3) as a result of the foregoing, Digimarc's subscription revenue and annual recurring revenue would be adversely affected; and (4) as a result of the foregoing, defendants' positive statements about Digimarc's business, operations, and prospects were materially misleading and/or lacked a reasonable basis. When the true details entered the market, the lawsuit claims that investors suffered damages. To join the Digimarc class action, go to or call Phillip Kim, Esq. at 866-767-3653 or email [email protected] for more information. No Class Has Been Certified. Until a class is certified, you are not represented by counsel unless you retain one. You may select counsel of your choice. You may also remain an absent class member and do nothing at this point. An investor's ability to share in any potential future recovery is not dependent upon serving as lead plaintiff. Follow us for updates on LinkedIn: on Twitter: or on Facebook: Attorney Advertising. Prior results do not guarantee a similar outcome. ------------------------------- Contact Information: Laurence Rosen, Esq. Phillip Kim, Esq. The Rosen Law Firm, P.A. 275 Madison Avenue, 40th Floor New York, NY 10016 Tel: (212) 686-1060 Toll Free: (866) 767-3653 Fax: (212) 202-3827 [email protected] To view the source version of this press release, please visit
Associated Press
22 minutes ago
- Associated Press
ROSEN, GLOBAL INVESTOR COUNSEL, Encourages Tempus AI, Inc. Investors to Inquire About Securities Class Action Investigation
New York, New York--(Newsfile Corp. - June 6, 2025) - WHY: Rosen Law Firm, a global investor rights law firm, announces an investigation of potential securities claims on behalf of shareholders of Tempus AI, Inc. (NASDAQ: TEM) resulting from allegations that Tempus AI may have issued materially misleading business information to the investing public. SO WHAT: If you purchased Tempus AI securities you may be entitled to compensation without payment of any out of pocket fees or costs through a contingency fee arrangement. The Rosen Law Firm is preparing a class action seeking recovery of investor losses. WHAT TO DO NEXT: To join the prospective class action, go to or call Phillip Kim, Esq. toll-free at 866-767-3653 or email [email protected] for information on the class action. WHAT IS THIS ABOUT: On May 28, 2025, before the market opened, published an article entitled 'Tempus AI stock sinks following Spruce Point short report.' The article stated Tempus AI shares had fallen after 'the company was targeted in a short-seller report by Spruce Point. The report raised serious concerns about the integrity of Tempus AI's product, the credibility of its management, and its financial reporting practices.' On this news, Tempus AI stock fell 19.2% on May 28, 2025. WHY ROSEN LAW: We encourage investors to select qualified counsel with a track record of success in leadership roles. Often, firms issuing notices do not have comparable experience, resources, or any meaningful peer recognition. Many of these firms do not actually litigate securities class actions. Be wise in selecting counsel. The Rosen Law Firm represents investors throughout the globe, concentrating its practice in securities class actions and shareholder derivative litigation. Rosen Law Firm achieved the largest ever securities class action settlement against a Chinese Company at the time. At the time Rosen Law Firm was Ranked No. 1 by ISS Securities Class Action Services for number of securities class action settlements in 2017. The firm has been ranked in the top 4 each year since 2013 and has recovered hundreds of millions of dollars for investors. In 2019 alone the firm secured over $438 million for investors. In 2020, founding partner Laurence Rosen was named by law360 as a Titan of Plaintiffs' Bar. Many of the firm's attorneys have been recognized by Lawdragon and Super Lawyers. Follow us for updates on LinkedIn: on Twitter: or on Facebook: Attorney Advertising. Prior results do not guarantee a similar outcome. ------------------------------- Contact Information: Laurence Rosen, Esq. Phillip Kim, Esq. The Rosen Law Firm, P.A. 275 Madison Avenue, 40th Floor New York, NY 10016 Tel: (212) 686-1060 Toll Free: (866) 767-3653 Fax: (212) 202-3827 [email protected] To view the source version of this press release, please visit
Associated Press
22 minutes ago
- Associated Press
Imaginarium Labs Announces Launch of AI-Assisted Visual Writing Platform for Storytellers
Imaginarium Labs announces the launch of its all-in-one writing platform that enhances storytelling through AI-powered structure, visual planning, and creative control. United States, June 7, 2025 -- Imaginarium Launches Comprehensive AI-Enhanced Writing Platform Focused on Visual Story Development Imaginarium Labs has officially announced the release of its AI-assisted visual writing platform, marking a significant innovation in digital storytelling. Designed to transform the traditional writing experience, the new platform introduces a human-centered interface that blends creative autonomy with AI-enhanced planning and narrative structure. The announcement follows a three-year development cycle led by founder Alex Stanin and a focused team of engineers, designers, and AI specialists. Their mission was to build a product that allows storytellers to manage complex narratives, develop immersive worlds, and streamline their creative process—all without compromising artistic integrity. Stanin stated that the platform was born out of a personal struggle with fragmented writing tools and a desire to craft a single environment where imagination could flourish without distraction. Announcement of a New Creative Paradigm The official launch of the introduces what the company calls a 'visual narrative universe' for writers, authors, screenwriters, and collaborative teams. The platform provides a seamless space for outlining, drafting, character development, and formatting—offering visual tools to organize timelines, track themes, map relationships, and explore world-building with clarity. The announcement addresses a key challenge in modern storytelling: managing increasingly complex projects while maintaining creative momentum. The Imaginarium app offers a structured yet flexible approach, enabling users to plan and revise with the benefit of spatial design elements rarely found in writing platforms. The company confirmed that the Imaginarium App is now fully operational and open to the public. The tool includes dedicated modules for fiction authors, screenwriters, and multi-format creators developing content for television or digital media. AI-Assisted Storytelling With Human Control In response to growing debates around AI-generated content, Imaginarium announces its commitment to a writer-first AI model. The platform uses artificial intelligence to offer intelligent suggestions on pacing, dialogue, character consistency, and world logic—but it does not write the story. According to Stanin, the role of AI in Imaginarium is to reduce creative fatigue, not author the work. 'Our mission is simple,' said Stanin. 'We want writers to stay in full control of their narratives. The AI handles structure and repetitive tasks so creators can focus on storytelling.' The AI co-pilot is presented as an optional tool rather than a driver, reinforcing the company's position that storytelling is a human endeavor enhanced by technology, not overridden by it. All-in-One Toolkit for Novelists, Screenwriters, and Creative Teams With the official release, Imaginarium becomes one of the first platforms to integrate a full suite of creative tools under one architecture. This includes chapter planning, auto-formatting for screenplays, print-ready manuscript exporting, detailed location and character builders, and real-time team collaboration. Unlike conventional tools that specialize in one area of writing, Imaginarium announces support for the entire creative lifecycle—from ideation to final manuscript formatting. The platform's visual structure enables authors to view and interact with their content spatially, making it easier to identify narrative gaps or inconsistencies. The company has confirmed support for individual creators and collaborative teams, positioning Imaginarium as scalable for studios, co-writers, and development houses managing serialized content or franchise universes. Commitment to Ethical Technology and Creative Empowerment In its public launch announcement, Imaginarium Labs emphasized its ethical approach to artificial intelligence, prioritizing creative integrity over automation. All AI components within the platform are designed to assist rather than dictate, focusing on consistency checks, dialogue flow, and pacing improvements. The platform also distinguishes itself through its visual design philosophy. Story elements are not confined to text blocks or folders but are presented in interconnected maps, timelines, and networks that mirror how many creators think and plan. This approach is intended to foster deeper engagement with story mechanics and narrative cohesion. Strategic Development and Future Vision The company has announced ongoing investment in community engagement and educational resources. Plans include webinars, storytelling forums, interactive tutorials, and mentorship opportunities that connect users to industry professionals. Stanin confirmed that the company is actively gathering feedback from early adopters to inform feature enhancements and roadmap priorities. The goal is to evolve Imaginarium into a central creative hub for writers across mediums and skill levels. The launch announcement also included a call to action for writers looking to join a growing ecosystem committed to elevating their work through thoughtful design and intelligent support tools. About Imaginarium Imaginarium is a next-generation writing platform that merges visual planning, narrative structure, and AI-powered support to assist storytellers across genres and formats. Founded in 2022 by Alex Stanin, Imaginarium's mission is to provide creators with a distraction-free environment that promotes deep storytelling, collaborative development, and intuitive control over complex narratives. The platform is designed to serve novelists, screenwriters, world-builders, and creative teams seeking an integrated solution for story development. Media Contact Name: Alex Stanin Email: [email protected] Title: Founder Contact Info: Name: Alex Stanin Email: Send Email Organization: Imaginarium Website: Release ID: 89161858 Should any problems, inaccuracies, or doubts arise from the content contained within this press release, we kindly request that you inform us immediately by contacting [email protected] (it is important to note that this email is the authorized channel for such matters, sending multiple emails to multiple addresses does not necessarily help expedite your request). Our dedicated team will promptly address your concerns within 8 hours, taking necessary steps to rectify identified issues or assist with the removal process. Providing accurate and dependable information is at the core of our commitment to our readers.
