Meta Platforms is actively developing an artificial intelligence version of CEO Mark Zuckerberg. This AI avatar, currently in its early stages, aims to enhance Zuckerberg's communication with employees and is part of Meta's broader investment in personal superintelligence. The AI is being trained on Zuckerberg's public statements and communications. However, Meta, along with OpenAI and ByteDance, is also identified by Akamai as a top culprit in a 300% surge of AI bot activity targeting publishers in 2025, raising concerns about content consumption without benefit to creators.
In the realm of AI coding, tools like Cursor, Claude Code, and OpenAI Codex are forming a new composable stack, focusing on specialized layers for orchestration, execution, and review rather than merging. Cursor 3, for instance, introduces an Agents Window, while an OpenAI plugin integrates Codex directly into Claude Code. Meanwhile, Anthropic's Claude Mythos Preview model revealed significant cybersecurity risks during lab testing, demonstrating powerful vulnerability exploitation skills. This led Anthropic to withhold its public release, sharing it only with partners for cybersecurity purposes and prompting the formation of Project Glasswing to address these threats. On the regulatory front, the FDA denied Harrison.ai's petition to exempt certain AI radiology devices from premarket review, citing patient safety concerns, a stance supported by the American College of Radiology.
Autonomous AI agents are reshaping e-commerce, with 25-40% of users in developed markets already relying on AI for product discovery. This shift requires businesses to optimize for algorithmic attention through structured metadata and API-first product feeds. Addressing the security of these agents, Cisco plans to acquire Astrix Security, an Israeli startup specializing in managing and securing AI agents within organizations by mapping them and implementing just-in-time access policies. In healthcare, major insurance companies are using AI for coverage decisions, but this practice faces class action lawsuits alleging wrongful treatment denial, with experts warning that AI trained on flawed human data could perpetuate existing issues, posing significant risks to patients. The ET AI Most Innovative AI Product Awards 2026 will honor AI products from startups, SMEs, and enterprises in India this June.
Key Takeaways
- Meta is developing an AI version of CEO Mark Zuckerberg to facilitate employee interaction, trained on his public statements.
- OpenAI, Meta, and ByteDance are identified as top sources of a 300% surge in AI bot activity targeting publishers in 2025, consuming content without benefiting creators.
- The FDA denied Harrison.ai's petition to exempt certain AI radiology devices from premarket review, citing patient safety concerns.
- AI coding tools like Cursor, Claude Code, and OpenAI Codex are forming a new composable stack, emphasizing specialized layers for orchestration, execution, and review.
- Anthropic's Claude Mythos Preview model, capable of exploiting software vulnerabilities, was not publicly released due to cybersecurity risks, leading to the formation of Project Glasswing.
- Agentic AI is transforming e-commerce, with 25-40% of users relying on AI for product discovery, requiring businesses to optimize for algorithmic attention.
- Cisco plans to acquire Astrix Security to enhance its AI security portfolio by managing and securing autonomous AI agents within organizations.
- Health insurance companies are using AI for coverage decisions, leading to class action lawsuits alleging wrongful treatment denial and raising patient safety concerns.
- The ET AI Most Innovative AI Product Awards 2026 will recognize innovative AI products from startups, SMEs, and enterprises in India across three tracks this June.
Meta developing AI Mark Zuckerberg for employee interaction
Meta Platforms is creating an artificial intelligence version of CEO Mark Zuckerberg to interact with employees. This AI avatar is still in its early stages and aims to help Zuckerberg communicate more effectively with Meta's workforce. The development is part of a broader effort by Meta to advance its personal superintelligence capabilities. The AI is being trained on Zuckerberg's public statements and communications.
Meta trains AI Mark Zuckerberg for employee interaction
Meta is developing an AI version of CEO Mark Zuckerberg to help employees interact with the technology. This AI persona is being trained on his public statements and communications. The company is also working on AI systems that can act as personal assistants. This initiative is part of Meta's significant investment in AI research and development, aiming to power various products and services.
FDA rejects AI radiology device review exemption
The U.S. Food and Drug Administration (FDA) has denied a petition to exempt certain AI radiology devices from premarket review. The petition, submitted by Australia-based AI developer Harrison.ai, sought to bypass the standard review process. Many industry leaders, including the American College of Radiology, expressed concerns about patient safety. The FDA sided with safety concerns, noting that most of the 45 comments received opposed the exemption.
AI coding tools Cursor Claude and Codex form new stack
AI coding tools Cursor, Claude Code, and OpenAI Codex are forming a new composable AI coding stack instead of merging into one product. This approach splits specialized layers for orchestration, execution, and review, similar to how developers use multiple infrastructure tools. Cursor 3 introduces an Agents Window for managing multiple AI agents, while an OpenAI plugin allows Codex to integrate directly into Claude Code for reviews and task delegation. This layering strategy allows tools to work together rather than consolidate.
Anthropic's Mythos AI reveals cybersecurity risks
Anthropic's new AI model, Claude Mythos Preview, has uncovered significant cybersecurity risks during lab testing, leading to its decision not to be publicly released yet. The AI demonstrated powerful cybersecurity skills, capable of finding and exploiting software vulnerabilities. Anthropic has shared access to Mythos with partner organizations for cybersecurity purposes only. This situation has prompted the formation of Project Glasswing, involving major tech companies to address potential cybersecurity threats.
AI agents are changing e-commerce discovery and sales
Agentic AI, using autonomous shopping agents, is shifting e-commerce from human-driven discovery to algorithmic decision-making. Between 25-40% of users in developed markets already rely on AI tools for product discovery and comparison. This requires businesses to optimize for algorithm attention by providing structured metadata, real-time pricing, and trust signals. Platforms and brands must adapt by offering API-first product feeds and revised monetization strategies to cater to these AI agents.
Publishers face surge in AI bot and scraper activity
Publishers are increasingly targeted by a growing ecosystem of AI bots and third-party scrapers, making content protection difficult. Akamai reported a 300% surge in AI bot activity in 2025, with media, particularly publishers, being significant targets. OpenAI, Meta, and ByteDance were top culprits. These AI fetchers and training crawlers consume content and infrastructure without benefiting publishers, potentially leading to declining referral traffic and revenue.
ET AI Awards 2026 seeks innovative products
The ET AI Most Innovative AI Product Awards 2026 will honor AI products from startups, SMEs, and enterprises in India this June. The awards feature three distinct tracks to evaluate products within their appropriate context of scale and innovation. The Startup track focuses on novel approaches and future potential, the SME track emphasizes practical application and measurable business impact, and the Enterprise track highlights scalability and strategic outcomes. Each nominee will receive a digital profile for increased visibility.
Cisco plans to acquire Astrix for AI agent security
Cisco intends to acquire Astrix Security, an Israeli startup focused on managing and securing autonomous AI agents within organizations. Astrix's platform maps AI agents, identifies risks like unauthorized access or excessive permissions, and implements security measures such as just-in-time access policies. The acquisition aims to enhance Cisco's AI security portfolio by helping companies manage AI agent risks and streamline development processes.
AI in health coverage raises patient risks
Major health insurance companies are increasingly using artificial intelligence to make coverage decisions, aiming to save costs. However, class action lawsuits allege that insurers are wrongfully withholding treatment using AI. Experts warn that AI trained on data from flawed human systems could perpetuate or worsen these issues. While AI offers potential benefits, its use in healthcare coverage decisions carries significant risks for patients.
Sources
- Meta builds AI version of Zuckerberg to interact with employees- FT
- Meta builds AI version of Mark Zuckerberg to interact with staff
- FDA denies petition to exempt certain radiology AI devices from premarket review
- Cursor, Claude Code, and Codex are merging into one AI coding stack nobody planned
- Anthropic Mythos Reveals Pandora’s Box Of AI Extensional Risks And For Safety Sakes Not Yet Publicly Released
- AI Agents Reshape E-commerce Discovery and Sales
- In Graphic Detail: New data shows publishers face growing AI bot, third-party scraper activity
- ET AI Most Innovative AI Product Awards 2026: Startups, SMEs, Enterprises: Where does your product fit?
- Cisco aims to make AI agents safer with proposed acquisition of Astrix
- Watch: As AI Makes More Health Coverage Decisions, The Risks To Patients Grow
Comments
Please log in to post a comment.