Anthropic's Fable shutdown highlights AI backup plan need

Key Takeaways

AI Backup Plans: A Must for CX Leaders

Anthropic's shutdown of Fable highlights the need for CX leaders to have a backup plan for AI models. Ashish Nagar, CEO of Level AI, emphasizes the importance of control over AI systems and not being fully dependent on third-party providers. Enterprises using AI in critical workflows need more control over the systems they rely on. Nagar suggests that buyers should look beyond headline model performance and ask tougher procurement questions. AI is increasingly being used in customer experience operations, making it crucial to have a backup plan.

The AI Identity Gap in Customer Experience

Geoffrey Mattson, CEO of SecureAuth, discusses the need for stronger identity controls in AI agents. Anthropic's shutdown of Fable underlined the challenge of verifying who is using AI models. Traditional login details offer little assurance when autonomous systems are acting through application programming interfaces and delegated workflows. Enterprises must understand who owns an agent, what systems it can access, and when a human should approve higher-risk actions.

The Five Pillars of Agentic AI Security

Agentic AI security requires a governance framework that manages five key issues: policy-as-code, least-privilege identity for agents, human oversight at defined thresholds, end-to-end traceability, and continuous risk monitoring. Enterprises must prioritize security and oversight as agentic systems move from pilot to production. Regulatory expectations are catching up fast, and boards are starting to ask pointed questions about accountability.

AI Cameras Track Gray Whale Behavior in Bay Area

Researchers in the Bay Area are using AI cameras to track gray whale behavior. The cameras detect heat changes generated when whales surface and relay sightings to vessels on the Bay. The system aims to help captains avoid deadly collisions and provide researchers with valuable insights into whale behavior. The cameras have already detected gray whales daily for 48 days straight.

Pentagon's AI Strategy: Speed and Caution

The Pentagon's AI strategy emphasizes speed in automating non-mission-critical back-office functions, but caution is required in mission-critical, lethal applications. AI-based fires should proceed with speed, but in an offensive kill chain, reliability, accountability, and human judgment are crucial. The Pentagon is investing in more compute hardware for AI and deploying models on classified networks.

HHS Update on Artificial Intelligence RFI

The Department of Health and Human Services (HHS) is accelerating the adoption of artificial intelligence (AI) in clinical care settings. HHS issued a Request for Information (RFI) on AI tools to deflate healthcare costs and improve patient outcomes. The department received over 7,000 comments and is working on implementing AI tools within a strict regulatory framework.

Conference Organizers Face Criticism for Hidden Prompts

Scientists are criticizing a conference for using hidden prompts to catch peer reviewers using generative AI. The conference organizers added concealed instructions for large language models to use telltale phrases in peer-review reports. Some researchers argue that this approach erodes trust and may filter out some problematic submissions but won't solve bigger problems with peer review.

Sodexo Embraces AI and Robotics in the Kitchen

Foodservice giant Sodexo is embracing AI and robotics to reshape the kitchen. The company is using AI tools to automate menu planning, ingredient management, and kitchen operations. Sodexo's chief tech, data, and digital officer, Alice Guehennec, says AI is being used to increase adoption and put AI everywhere at scale.

Japan's $6 Billion AI Investment

Japan plans to invest around $6 billion in a homegrown artificial intelligence model and have 10 million AI-equipped robots operating in various sectors by 2040. The government aims to develop a sovereign AI model to reduce reliance on technology from the US and China. The investment will focus on physical AI and robots to help plug gaps in the workforce.

Judiciary Adds AI and Disability Law Training

The Supreme Court of Korea has added new courses to its mandatory training program for judges, focusing on AI and disability law. The new courses aim to enhance judges' professionalism and competence in handling complex cases involving AI and disability issues. The judiciary is also establishing a center for AI and disability law to provide research and training.

Sources

• Anthropic’s Fable Shutdown Proves CX Leaders Need an AI Backup Plan
• Anthropic’s AI Identity Gap: Why CX Leaders Need Agent Controls
• The 5 pillars of agentic AI security
• New AI cameras providing Bay Area researchers insight into gray whale behavior
• Speed where it’s safe, caution where it might kill: How the Pentagon should use AI
• HHS Provides Update on its Artificial Intelligence RFI
• Scientists decry conference’s use of hidden prompts to snare AI peer reviews
• How foodservice giant Sodexo is embracing AI and robotics to reshape the kitchen
• Japan To Invest $6 Billion In AI, Have 10 Million AI Robots By 2040
• Judiciary Adds AI, Disability Law, Courtroom Conduct to Mandatory Training