ASI (Fetch.ai) news

A coordinated attack on Sept. 19 drained 8.7 million FET (about $1.55 million) from SingularityNET's Ethereum bridge and used a compromised NuNet minter key to mint 408.5 million NTX, sent to the same wallet. Investigators linked the theft to compromised backend authorization credentials, not a contract flaw, and noted the contract's transaction cap was not enforced on conversionIn. The NTX mint, created 29 minutes after the FET withdrawal, equaled roughly 42% of NuNet's supply. The attacker exchanged stolen assets via MetaMask, holding 547.89 ETH ($1.44 million) and 230 million NTX by Sept. 20. Fetch.ai paused AGIX-to-FET conversions and the bridge as a precaution, while Bitvavo suspended WMTX deposits after a separate security incident. Credentials remained unrotated hours after the attack, underscoring the need for remediation.

A single exploiter has been linked to attacks on Fetch.ai and NuNet, draining approximately $2 million in assets. Security firm Blockaid reported that the attacker stole about $1.56 million worth of FET from a Fetch.ai token converter on Ethereum using a valid signature. The same wallet also received a large NTX mint from the NuNet deployer account, valued at roughly $452,000. The exploit caused NuNet's NTX token to plunge over 70% in 24 hours, hitting an all-time low of $0.000328. Fetch.ai's FET token fell about 5% amid a broader market selloff. The combined incidents push September's total crypto losses past $333 million, according to DefiLlama, which had already logged $331 million in losses earlier this month.

Fetch.ai has released the A2A Outbound Adapter, a tool enabling its autonomous agents to communicate with external AI agents using the Google-linked Agent-to-Agent (A2A) Protocol. The adapter, part of the uAgents-adapter package, offers two configuration modes: SingleA2AAdapter for one server and MultiA2AAdapter for multiple servers, with features including chat protocol forwarding, agent discovery, health monitoring, and keyword matching. It includes a payment bridging mechanism between A2A/AP2 mandates and Fetch.ai's Payment Protocol, with a demonstrated cart store example using Skyfire payments. The adapter requires Python 3.10 or higher and is available in versions 0.6.0 and 0.6.2. It was incrementally developed across 2025-2026, ahead of the anticipated A2A Protocol v1.0.0 release in March 2026, and is not tied to new token mechanisms or staking incentives.

Fetch.ai's Innovation Lab has deployed custom AI agents at US and UK universities, including San José State University and California State University, Long Beach, to help new students navigate campus and discover events. The agents, built on Fetch.ai's uAgents framework, Agentverse, and ASI:One, include a Welcome Week AI Agent and a gamified Scavenger Hunt bot. The initiative is part of a broader educational strategy expanding through 2026, involving workshops, hackathons, and a $100,000 grant to UCLA. By engaging computer science students, Fetch.ai aims to build a pipeline of developers familiar with autonomous agent technology. Notably, these campus deployments do not explicitly integrate blockchain or crypto tokens, functioning as straightforward AI tools.

Fetch.ai is expanding access to Athena, its AI execution layer, by offering three free projects to every ASI:One platform account, a move announced on September 5, 2026. Athena, which launched exclusively for PRO subscribers on August 13, is designed for complex, multi-step tasks beyond standard chatbot interactions, using over 50 specialized tools for financial modeling, data analysis, research, and report generation. The free access strategy aims to drive adoption by letting users experience the difference between simple Q&A and delegating full projects. Unlike a standalone model, Athena integrates reasoning, planning, and agent coordination through the Agentverse, orchestrating multiple AI agents for sustained work.

AI-linked crypto tokens surged after Nvidia CEO Jensen Huang outlined the company's AI infrastructure vision at the GTC keynote. NEAR rose over 10% to its strongest level since January, FET spiked up to 20% intraday, WLD gained about 10% near $0.40, and GRASS jumped 13% to 2026 highs. Huang projected roughly $1 trillion of chip demand backlog through 2027, with hyperscale cloud providers accounting for 60% of Nvidia's business, and highlighted the rapid rise of agentic AI, citing OpenClaw and the enterprise‑ready NemoClaw. While he did not mention crypto, blockchain projects are positioning themselves to provide rails for autonomous AI agents. Nvidia shares closed about 1.5% higher.

Fetch.ai announced that its autonomous AI agents will soon be able to complete payments on users’ behalf. The system, set for a January 2026 rollout, uses Visa’s infrastructure to issue single‑use card credentials tied to each transaction, eliminating permanent numbers and adding KYC and an identity layer so agents act transparently on behalf of identifiable users. In live tests the agents have booked restaurant reservations and paid deposits, and the rollout is delayed for additional Visa checks; Mastercard support will follow. Fetch.ai partners with established financial providers, avoids direct bank transfers, and focuses on a model where users can host and manage their own agents, aiming to transition the web‑based economy to an AI‑first one.

Fetch.ai and three token holders sued Ocean Protocol in the Southern District of New York on Nov. 4 2025, alleging fraud and misrepresentation of the OceanDAO’s autonomy. The complaint says Ocean moved ~700 million OCEAN community tokens to a Cayman entity, converted ~661 million OCEAN into 286 million FET, and sold ~263 million FET, breaching promised community rewards and harming FET prices. Ocean disputes the claims, describing the suit as a social‑media narrative, while the case highlights challenges in DAO governance and token stewardship.