On June 2, 2026 President Donald Trump signed an executive order establishing a voluntary federal framework for reviewing the most advanced artificial intelligence models before public release. The order asks developers to give U.S. agencies access to “covered frontier” models for up to 30 days for cybersecurity testing and evaluation, a compromise from an earlier 90-day draft. It directs Treasury, the NSA, CISA, DHS and other agencies to develop classified benchmarks to identify models with advanced cyber capabilities and to stand up an AI cybersecurity clearinghouse to coordinate vulnerability discovery and remediation with industry. The order also asks the Justice Department to prioritise AI-assisted cybercrime enforcement. White House language explicitly bars the creation of mandatory licensing, preclearance or permitting regimes, reflecting industry pressure that scuttled an earlier, tougher version of the measure. The move was partly prompted by Anthropic’s Mythos model and follows weeks of internal administration debate and late-stage changes to the policy. Reactions were mixed: major firms expressed cautious support while safety advocates warned the voluntary approach may lack sufficient enforcement.

On June 3, 2026 Britain’s Competition and Markets Authority (CMA) imposed new conduct requirements on Google’s search services, giving online publishers the right to opt out of having their content used in Google’s AI Search features and for training/fine‑tuning its models. The CMA, which designated Google as holding strategic market status in general search, said the measures will also require clearer attribution (direct links) in AI‑generated summaries, transparency on how content is used, and new publisher metrics. Google said it will begin testing a control in Search Console that lets site owners manage how links and content appear in AI Overviews, AI Mode and Discover, and will roll the tools out to a subset of UK publishers before wider deployment. The regulator has given Google nine months to implement the package but expects key elements earlier. The CMA framed the move as a “world‑first” to rebalance bargaining power after publishers reported steep traffic losses from AI summaries; Google said opting out will not affect ordinary search rankings.

Apple will begin enforcing age verification for new Apple Accounts in Texas on June 4, 2026, to comply with state law SB 2420 after a federal appeals court temporarily stayed an injunction that had blocked the measure. New users in Texas will be asked to confirm they are 18 or older using a credit card, government ID or automated checks such as account age and payment methods on file; existing accounts are not affected. Users under 18 must join Family Sharing and obtain parental or guardian consent for downloads, in‑app purchases and “significant” app updates. Developers distributing apps in Texas must implement Apple’s Declared Age Range and Significant Change APIs, support parental consent revocation and can face civil penalties if they fail to comply. Apple had lobbied against the law and warned about data‑collection and privacy tradeoffs; Google’s Play Store is also subject to the statute. Legal challenges continue and courts will still decide the constitutionality and scope of SB 2420 while the state enforces it.

Google Labs on June 3 launched Dreambeans, an experimental AI app for iOS and Android that auto-generates a limited daily set of illustrated “stories” drawn from a user’s Google data. Using Personal Intelligence and Google’s Nano Banana 2 image model, Dreambeans connects — with user permission — to Gmail, Calendar, Photos, YouTube and Search history to curate 10–14 personalized lifestyle suggestions each morning. Stories range from local venue recommendations and trip planning to practical how-tos (for example, puppy training) and links to buy tickets or book services. Illustrations can incorporate users’ Google Photos face groups. The app is available only to U.S.-based Google AI Ultra subscribers aged 18+ (the $100-per-month tier) while other account holders can join a waitlist. Google says users can choose which services to connect, delete Dreambeans data, and give feedback to refine recommendations. Observers note the app’s potential to reduce endless scrolling but also highlight the broad scope of data access detailed in app privacy disclosures.

Amazon on June 3 began surfacing AI-generated product images inside the Amazon Shopping app as users type search queries, the company said. The synthetic images — shown beneath autocomplete suggestions — are intended to help shoppers who can visualise an item but not name it (examples Amazon gives include styles such as “cowl neck” or materials like “rattan”). The feature currently works for apparel and home goods on iOS and Android for U.S. customers; the pictured items are not actual listings but tapping a generated image directs users to search results for similar real products. The rollout is part of a wider push to embed generative AI across Amazon’s commerce experience, alongside shop-by-style collages, Lens Live visual search, Alexa for Shopping and AI review summaries. Amazon’s timing precedes Prime Day (June 23–26). Critics and reporters warn the fake images could mislead shoppers or create disappointment if users expect the depicted item to be sold on the site.

Microsoft has warned that starting July 13, 2026, Office apps on older Apple devices will enter a “reduced functionality mode” that prevents creating, editing or saving files — allowing only viewing and printing. The change affects Microsoft 365 subscribers and standalone versions including Office 2019 and Office 2021 on macOS, iPhone and iPad when devices run unsupported operating systems. Microsoft says macOS 12 Monterey, iOS 17 and iPadOS 17 will be the minimum supported versions; devices that cannot upgrade will lose editing capabilities. Microsoft attributes part of the disruption to an expiring security certificate: the company renewed the certificate and updated newer Office releases to recognise it but did not provide an update path for Office 2019 for Mac, which will effectively become read-only. Affected users are advised to update device OS and Office apps, use the free web versions of Microsoft 365, or subscribe to Microsoft 365. Microsoft says it will email affected customers who cannot upgrade. Windows Office versions are not impacted by the certificate issue.