"Google and Apple are integrating agentic capabilities like Gemini Intelligence with Google Gemini and Onscreen Awareness with Siri AI deep into their operating systems," wrote Meredith Whittaker in Signal's blog post.
Signal's screenshot defense and the decision to block agents
Signal has updated its iOS and Android apps to block AI agents such as Google Gemini and Apple Siri from taking screenshots of messages unless users explicitly opt in. Signal CEO Meredith Whittaker said the controls — which by default block Siri on iOS and Gemini and other agents on Android — are designed to stop agentic software from "scanning" the screen and capturing private communication that the agents then process with optical character recognition (OCR).
Whittaker tied the change to a prior protection Signal implemented last year to defend against Microsoft Recall, and she framed the move as necessary because "in doing so, your Signal data is captured, then made available to these agentic systems in order to allow them to answer your questions and do your bidding." Signal insists users who wish to take screenshots can disable the protection via the app’s settings, but warns that doing so "risks exposing your Signal content (and that sent by your contacts) to AI agents."
Partial fixes and file-system exposure
Signal acknowledges that blocking screenshots is only a partial solution. Whittaker pointed out that "AI agents may have other ways to access application data through the device file system." She also criticized platform vendors for not providing app developers with robust, simple mechanisms to restrict OS-level AI access without degrading user experience, and said this shortcoming is why Signal has been unable to implement the same protections on macOS.

This site is the portfolio.
OSINTSights runs on Cloudflare Workers, D1, R2, and Vectorize, with an AI pipeline on Hetzner ARM. Nubivance designed, built, and operates it. We do the same for clients.
See what we buildApple's macOS 27.3 changes and the permission trade-offs
Apple is introducing more granular controls for macOS with 27.3, currently in public beta testing. The changes affect two APIs: Device Control and Data Access, and Full Disk Access. Apple says these updates will produce "more permission prompts for users of macOS apps that seek access to user data," and will require developers of backup software and assistive technologies that use Accessibility APIs to apply for an entitlement from Apple to obtain Full Disk Access.
When announcing the Full Disk Access change, Apple warned that "as AI agents become increasingly capable and autonomous, the risks associated with this level of access will grow substantially," and framed the update as a way to help users "make informed decisions about their own data and privacy." The company is also reportedly working on its own AI agent to compete with Anthropic and OpenAI, raising the question — noted in the source material — whether Apple will use its operating system ownership as competitive leverage as it has in other platform fights.
Wider marketplace reaction: blocking, inviting, and detecting agents
The Signal decision comes amid a broader industry pattern of companies erecting limits around third‑party AI agents. Amazon has blocked shopping agents like Meta Muse from its e-commerce site (though not from AWS Marketplace); Walmart has invited third‑party agents to interact through its AI Gateway. Anthropic has blocked subscription customers from using third‑party harnesses. United Airlines has banned transactions from unauthorized AI bots. Similar barriers or limitations have been imposed by Target, Home Depot, Etsy, eBay, JPMorgan Chase, Charles Schwab, Citigroup, and Wells Fargo.
Concurrently, a bot-detection and mitigation sector has been growing, led by firms such as Cloudflare, Akamai, and DataDome, to detect and limit automated agents that interact with web services. The source argues that the web-browser-era notion of a "user agent" serving the user has blurred under data-collection and advertising incentives — and that AI agents now face comparable scrutiny about whom they ultimately serve.
What this means for technologists, policymakers, and end users
- Technologists and security teams: Developers lack "robust tools" from platform vendors to prevent agentic access without harming user experience, Signal warns. Teams will be asked to detect screenshots and other agentic behaviors and to plan for file-system vectors that screenshot-blocking does not stop.
- Policymakers and platform governance: Apple's macOS 27.3 shows an OS-level push toward finer-grained permissions and entitlements for Full Disk Access. Regulators and platform policy teams will need to weigh user transparency against developer burden when authorizing entitlements for backup and assistive‑tech vendors.
- End users and consumers: Signal's default protections reduce the risk that conversational content is swept into third‑party AI systems, but the company stresses that opting out of those protections exposes not only the user’s messages but also content sent by their contacts.
Signal’s move crystallizes a tension at the center of modern platforms: agents promise convenience and context, yet when those agents "spirit data away to the cloud" they can displace the direct relationship between people and services and expose private content to third-party servers. Signal has chosen to lock the door by default; the door remains unlockable, and the lock imperfect — while Apple’s nascent macOS controls and sweeping commercial responses show the market and platforms are still figuring out where the dams should be built.




