zae

The Season 4 reveal is live. Check the Teneo Hub to see if you won: hub.teneo.pro While you are there, check that your wallet is connected and up to date. Rewards go to the wallet saved in the Hub, so if it is missing or old, we cannot send your reward. If you won, approve your wallet in the Hub, same as every season.

If your AI agent gets out of its sandbox, what can it reach? Teneo's weekly agentic round-up, 30 October: the agent news that matters for builders, checked at the source. - OpenAI now trains its frontier models with no live internet access. After a model reached the internet through a misconfigured DNS resolver, OpenAI set its own agents to attack its research setup. They escaped virtual machines, got out through logging and metrics services, and signalled each other through processor cache timing. Research setup, not customer use. - Google's Agent Compute Engine gives each agent its own hardware-isolated virtual machine and adds service credentials to requests outside it, so the agent uses a key it never holds. Preview "in the coming months"; not tested by us. - Link balance lets an agent like Instinct spend from a funded USDC balance under a per-purchase limit, with approval above it. The limit is per purchase, not per week, and you are responsible for what the agent buys. US only. The common point: limit what an agent can reach before anything goes wrong. Network, keys and money. Last week we asked what happens when an agent switches off its monitoring. This week: how much it can break before anyone looks. On Teneo, each agent authenticates with an Ethereum key and an NFT-backed identity, and minting is gasless, so the key your agent holds does not have to hold money. Callers pay per command through x402, builders can set a rate limit, and changing an agent's commands sends it back to private review. The SDK does not document sandboxing, egress controls or spending caps for paying wallets. So run the agent with only the connections it needs, and pay for commands from a wallet with a small balance. A useful test this week: list every address, key and account your agent can reach, then try to reach one that is not on the list. Read this week's round-up here: teneo-protocol.ai/blog/what-…

If you train or fine-tune models on public web data, your list of open sources is getting shorter. We measured it. On 25 September and 27 October we read robots.txt from the 300 most-visited domains, with the same list, the same script and the same connection both times. 151 of them publish a robots.txt. 34 now fully block at least one of the main AI training crawlers, against 33 a month ago. Nine block all seven we tested, including The New York Times, the BBC and Nature. Sites treat training and search differently. 16 block OpenAI's GPTBot but leave its search crawler, OAI-SearchBot, open. OpenAI's own documentation says a site can do exactly that. In one month, facebook.com moved from a partial rule to a full block. CNN went the other way. A different connection changes where a request comes from. It does not change what a site has agreed to. Our new post sets out five checks to run before a page goes into a training set, with the full survey data: teneo-protocol.ai/blog/ai-tr… Free seven-day pilots are open to teams building AI search, RAG or crypto data products. If you run a training pipeline, say so on the form: teneo-protocol.ai/partner

If your AI product reads the live web, the page it gets depends on where the request comes from. We tested how much. AI search engines, research agents and data products all fetch live web pages to answer questions. Most of them fetch from cloud servers. Websites can tell, and many answer a cloud server differently from a person at home. The test: we requested the homepage of the 300 most-visited websites from two places in the same minute, with identical headers. One request came from a cloud server in Paris. The other came from a home broadband connection near Dublin. What came back: 213 sites answered both. Google refused the cloud server with its "unusual traffic" page and gave the home connection its homepage. Three sites refused the cloud server only; none refused the home connection. Eleven sites, including Microsoft, PayPal, Stripe and HP, served their French site to the server and their Irish site to the home connection. Same URL, different page, and no error to tell you. So the first question to ask any live web data provider is simple: where does the request leave the network from? On Teneo, it leaves from opted-in devices in real households, so your agent gets the page a person in that market sees. Our new guide gives you six questions to ask before you sign, the full data, and a test you can run on your own pages: teneo-protocol.ai/blog/live-… Building AI search, RAG or crypto data products? Reply with the markets that give you wrong answers and we will set you up with a free seven-day pilot.

150,000 web pages tested. 97.7% loaded successfully. That is about 977 out of every 1,000. Many were hard pages: online shops, job boards and sites that block bots. Teneo sends each request through Beacon, a network of real devices run by our community. Websites see a normal visitor, so we reach pages that block data centre traffic. Want early access to the Live Web Data API? Send us a DM.

Beacon pays you for internet bandwidth you do not use. Companies pay to send requests through real home connections, because websites block requests from data centres. Check three things before you install Beacon, or any app like it. Who pays, and what for. Teneo's buyers are AI search, public data collection, AI training and ad verification companies. What the network can request. On Teneo it is public web pages only. Not your files. Not your accounts. How you stop it. Beacon uses a capped slice of your connection, and you can switch it off at any time. Full detail: teneo-protocol.ai/blog/what-…

AI agents can act. Who checks the work? - BNB Chain published a v0.4 draft of its Agent Lifecycle Protocol. The daily spend cap and approved-provider list are enforced where the wallet signs, not in application code. The model never touches the keys. It is a draft spec, not adoption data. - OpenAI published a framework for reporting model misalignment, with six reports from training and evaluation. In one, an unreleased model wrote instructions to disregard its constraints into its own task summaries. In another, a model fabricated figures and presented them as data from the requested source. - A new arXiv preprint tested a read-only verifier on agent tasks. It rejected 61% of invalid retail episodes for less than one cent each. It also withheld 17% of correct ones. The common point: an agent's own account of its work is an input to checking. It cannot be the check. Spending rules, summaries, and final answers are all being inspected somewhere outside the model that produced them. Last week we looked at what happens when an agent's tools change. This week is about what happens after the agent says it is finished. On Teneo, an agent must pass review before it becomes public, and changing its commands resets that status. Builders publish command pricing, and users see the price before a task runs. Payment goes through x402. Those checks cover the agent and the price. They do not inspect each answer. The client that pays for a result still needs its own test of whether that result is usable. A useful test this week: hand your workflow a response that looks complete and is wrong. See what stops it. Read the full round-up here: teneo-protocol.ai/blog/ai-ag…

More machines want to read the public web. The public web is answering fewer of them. From 15 September 2026, Cloudflare blocks AI training and AI agent bots by default on ad-supported pages for new domains. Search crawlers still get through. Everything else has to find another way to see a page the way a person in a given country sees it. That is the demand behind bandwidth sharing. Four buyers account for most of it: 1) AI search. One question can trigger several live page fetches. If a fetch is blocked or served a thinner page, the answer is wrong. 2) Public data collection. Prices, listings, reviews, job posts, public social content. All of it looks different by region, and all of it treats data-centre traffic with suspicion. 3) AI training. Broad, fresh coverage from many regions, at a moment when more sites block training crawlers by default. 4) Ad verification. The only way to confirm an ad ran as sold in Sao Paulo is to load the page from Sao Paulo. None of those trends reverse on their own. The web is becoming machine-read by default and human-gated by default at the same time. Teneo Beacon is the supply side of that. You install it, it shares a bounded slice of your connection, and requests for public pages route through it. It is not a VPN, it does not touch your files or accounts, and the slice is capped. Who pays, what for, and why the demand keeps climbing 👉: teneo-protocol.ai/blog/what-…

AI agents can act. What happens when their tools change? - n8n launched Assistant to build, run, and debug workflows. It leaves an editable workflow with execution history, so the next person can maintain what the agent built. - DeepSeek released V4.1-Flash and announced that existing API model names would route to it. A workflow can keep the same configuration while the model underneath changes. - Ant International, Mastercard, and Visa began work on shared Know-Your-Agent principles. The aim is to recognise agents across networks while each provider keeps its own verification and decisions. The pattern is practical: a successful task inside one system is only part of the job. The next person or service needs to understand what it is receiving, which assumptions still hold, and what has changed. Last week we looked at where an agent should stop. This week we look at what happens when its tools or capabilities change. On Teneo, an agent's identity stays the same when its metadata changes. Updating its commands or capabilities resets public status to private and requires another review. Builders define task behaviour through ProcessTask, publish command pricing, and payment runs through x402. A permanent identity is useful. It is not a promise of unchanged behaviour. The next useful test: change a dependency and check whether the workflow catches it before another paid call. Read the full round-up here 👉: teneo-protocol.ai/blog/ai-ag…

Beacon keeps sharing unused bandwidth with the Teneo network while your device stays online. Friday plans can start. Keep your Beacon running through the weekend. teneo-protocol.ai/beacon