Every published Top AI Stories item tagged with Perplexity AI, newest first.
Trellner Research queried two Perplexity models across 380 software categories and logged 7,534 citations, finding that 59.8 percent pointed to domains ranked below one hundred thousand on the Tranco list. Three sites sharing nameservers, page templates and registration windows — wifitalents, worldmetrics and gitnux — have produced 215,128 machine-generated buying guides between them and drew 181 of those citations. The authors are careful to say they have not shown the recommendations are wrong, only that much of the sourcing beneath them is manufactured.
Perplexity, working with Nvidia, brought its agentic Computer product to the desktop as Portable Computer, with local work consuming no credits. The engineering is the actual claim rather than the privacy pitch: the orchestrator running the loop is deterministic code instead of a second model, tool calls run in an operating-system sandbox, and the harness fails closed, disabling itself if that sandbox is unavailable. Escalating to a cloud model is permission-gated and shows the user what would leave the machine. It needs an Nvidia DGX Spark or a card with at least 24 gigabytes of video memory.
Responsible Statecraft reports that a firm called Piro, working for the Israeli Government Advertising Agency, created the Hanover Institute for Public Policy, which has published more than 100 reports since August 6 with footnotes, tables of contents and a neutral register. Piro's own site advertises "AI Story Optimization" and content "engineered for how LLMs evaluate credibility," and a co-founder posted about reverse-engineering how ChatGPT, Gemini and Perplexity assemble answers. GPTZero flagged eleven of twelve sampled articles as machine-written.
In an August 4 opinion, the Ninth Circuit vacated the injunction that had kept **Perplexity's Comet** browser off Amazon, holding that Amazon was unlikely to win its claim under the Computer Fraud and Abuse Act. The panel found it was the user, not Perplexity, who accessed Amazon's computers, because the assistant reads screenshots on the user's own device rather than talking to Amazon's servers. It is the first federal appellate ruling on whether AI agents may act on a user's behalf across third-party sites. The court called the result fact-specific and left Amazon's contract claims alive.
At Computex 2026, Perplexity unveiled what it calls the first hybrid local-cloud inference orchestrator for its Computer agent. A small model on the device sizes up each task and either runs it locally or, with the user's permission, routes it to a frontier model in the cloud — keeping financial, health, and personal files on-device by default. The model- and chip-agnostic system runs on Intel and NVIDIA hardware and is set to reach Perplexity Computer in July, with Windows support on a waitlist.
Perplexity launched Search as Code, an architecture that lets AI agents query its search stack by writing Python instead of calling fixed APIs. Agents can orchestrate primitives — parallel fan-out, deduplication, filtering, and ranking — before results ever reach the model's context, which Perplexity says cuts latency and delivers about 2.5-times the performance of standard tool-calling on complex retrieval. It is now the default in Perplexity's Computer product and is available through the company's Agent API.
Perplexity's autonomous Mac agent — previewed last month to a Max-tier waitlist — opened to all Pro and Max subscribers. Personal Computer can read local files, drive native Mac applications, orchestrate workflows across over 400 connectors, and be triggered remotely from iPhone. Tasks run in a sandboxed Perplexity-server environment. The general-availability move puts Perplexity in direct competition with Anthropic's Claude Computer Use and OpenAI's ChatGPT Operator — but with desktop file access as the differentiator.