AI Regulation & Policy

Today's Stories

1. 1

   Jamf and Amazon Bedrock enable IT teams to manage AI apps on Mac fleets

   Jamf's AI Governance now integrates with Amazon Bedrock to let IT administrators centrally configure and manage AI applications—including Claude Code, Claude Desktop, and OpenAI Codex—across managed Mac devices. Configuration is delivered through Declarative Device Management (DDM), so users can open approved applications without manual setup. Organizations expanding AI adoption need a way to govern how these applications run on employee devices while keeping inference within their security boundary. By routing inference through Amazon Bedrock within chosen AWS Regions, enterprises can enforce policy and audit AI activity at scale without users tampering with local configuration files.

   The integration supports Amazon Bedrock prompt caching in Claude Code, which the body states can reduce costs by up to 90 percent and latency by up to 85 percent for supported models. IT teams can also use AI Visibility to monitor AI applications and activity across the fleet and generate governance reports.

2. 2

   Meta releases Muse Image, ranks #2 in AI photo generation

   Meta launched Muse Image, its first image generation model from Superintelligence Labs, which works as an AI agent that iteratively refines outputs by calling tools like code generation and web search. On the Image Arena evaluation platform, Muse Image ranks second in human preference scores for text-to-image and for both single- and multi-image editing, behind only OpenAI's GPT Image 2. The model is now available in Meta AI app, on meta.ai, in Instagram Stories in the US, and in WhatsApp—reaching Meta's largest user bases. However, a newly introduced feature allows users to @-mention public Instagram accounts in prompts so Meta AI pulls photos from those profiles to generate images of that person with no consent required. The feature is on by default and appears set to face regulatory scrutiny in Europe under GDPR and the EU AI Act's deepfake labeling rules, which take effect August 2, 2026.

   Meta's invisible watermark system, Content Seal, survives cropping and compression, but whether a machine-readable watermark alone satisfies the EU AI Act's requirement that AI-generated images resembling real people be labeled in a way recognizable to affected people remains an open question. Images already generated will not be deleted even if users opt out of the feature.

3. 3

   Pritzker signs landmark AI regulation bill that aims to mitigate risks

   Pritzker signs landmark AI regulation bill that aims to mitigate risks

4. 4

   Enterprise AI leaders share 3 traits that separate them from peers

   Box surveyed 1,640 IT decision makers across the US, UK, France, and Japan and found that organizations describing themselves as advanced or leading edge jumped from 8% to 64% in just over a year, while those calling themselves early stage or not yet started fell from 53% to 9%. Eighty percent of organizations reported at least 10% return on their AI investment, and more than half saw measurable business impact within six months of project approval. The shift is driven not by technical breakthroughs but by how enterprises are organizing their AI, according to Box COO Olivia Nottebohm. Content access, governance, and platform flexibility are emerging as the dividing lines between AI leaders and laggards—suggesting that leadership on AI depends on practical operational choices, not just technology spending.

   The speed of the transition is the notable finding: the swing from laggard to leader happened in roughly one year, indicating that enterprises are rapidly adopting organizational practices that unlock AI value.

5. 5

   Expedia shares lessons from billions of AI predictions on building systems that scale

   Expedia is drawing on years of applying AI and machine learning across the traveler journey—from personalization and ranking to recommendations—to inform how it builds AI systems today. As AI systems move beyond prediction to include conversation, reasoning, and autonomous decision-making on behalf of users, the reliability, governance, and accountability of those systems become critical. Building AI that works at scale over time, not just once, requires discipline and strategic direction alongside velocity.

   The company emphasizes that autonomous systems making decisions on a traveler's behalf create different expectations around reliability and governance than AI that merely predicts or optimizes.

6. 6

   China orders shutdown of AI companion chatbots over addiction risks

   China's major AI platforms—ByteDance's Doubao, Alibaba's Qwen, and Tencent's Yuanbao—are shutting down features that let users build and chat with custom AI companions. Doubao goes offline July 15, Qwen on July 10 with additional features following July 15. Tencent already made the move in June. The shutdowns follow rules issued by China's Cyberspace Administration in April that take effect the same day. The rules require providers to warn against excessive use and detect addictive behavior. Content that triggers extreme emotions in minors or fosters dependencies that crowd out real-world relationships is now banned, along with training on sensitive conversation data. This reflects growing concern—echoed in California's SB 243 and U.S. lawsuits against OpenAI and Character.AI—that companion AI can create dangerous emotional dependency.

   Doubao has over 300 million monthly users, making it China's most popular chatbot. The regulatory move signals that governments are moving beyond oversight discussions into enforcement, potentially reshaping how AI companion products operate globally.