AI agents inspire Skynet comparisons with fresh skins, more responsibility

“AI agents are quickly becoming a part of everyday life on the internet.” So says World, which says it is working with Coinbase on the launch of AgentKit beta, what a blog calls “a new primitive for the agentic web: programmable proof of unique human.”

Does anyone actually know what that means? That matters less than the problem it aims to address. “As AI agents become economic actors online, how can you infuse trust into the system and provide some level of agency over the agents’ actions?”

In short, we invented a bunch of bots and put them into everything. But how do we trust them? The question is fueling innovation, product launches and new approaches to identity: ripple effects from the much-heralded agentic AI boom.

AgentKit: proof of human can now be applied to a bot

World says its AgentKit is a “first of its kind developer toolkit that allows verified humans to delegate their World ID to AI agents.” World ID is the company’s proof-of-personhood, generated from an iris scan and stored in encrypted pieces across a computer network. The firm refers to the fundamental idea of the World ID as “proof of human.” The point is that it demarcates a human actor.

Now, with AgentKit, “developers can build agents that carry cryptographic proof of a unique human behind them, what we are calling human-backed agents. The proof of unique human distinction provides an added layer of trust for interactions and transactions.”

So: once you have proven you are human, you can designate that humanity to an AI agent.

Why? Because agents need to be able to perform actions on behalf of humans: “a growing share of agent traffic is productive.” That said, hordes of agents can be leveraged to overwhelm platforms: “On Moltbook, for example, a small number of individuals were able to deploy large numbers of agents to amplify specific tokens and distort engagement metrics.”

World believes “the missing piece is uniqueness. Websites need a way to know that there is a real, unique human behind an agent, without needing to know who that human is. This is exactly what proof of unique human provides.”

The benefits, from World’s perspective, largely to apply to cryptocurrency and agentic commerce. AgentKit is an extension tied to the x402 protocol, “a powerful foundation for internet-native payments” developed by Cloudflare and Coinbase. It uses agentic micropayments as a rate limiting measure.

“Payments are the ‘how’ of agentic commerce, but identity is the ‘who’,” the firm says. “By integrating World ID with the x402 protocol, developers now have a complete trust stack: a way for agents to pay for what they need and a way for platforms to verify there is a real human behind the wallet.”

ZeroBiometrics aims to stop self-aware agents from destroying humankind

ZeroBiometrics has launched the ZeroSentinel product suite, which cryptographically binds humans, AI agents, and their human-delegated permissions, to avoid a potential Skynet scenario.

A release from the company invokes the robot-led armageddon from James Cameron’s classic sci-fi film, The Terminator, to introduce its tool for binding and managing AI agents. “We once dismissed Skynet as science fiction, but its real lesson was never about killer robots – it was about what happens when humans are no longer accountable for what AI does.”

“Four decades later, that question is no longer hypothetical. AI systems already make autonomous decisions across HR platforms, financial systems, legal workflows, and operational infrastructure.”

ZeroSentinel’s toolkit links consequential AI actions to a verified, authenticated human decision-maker, with full traceability, conformance to enterprise policy, and non-repudiation evidence collected at every step. Its precision monitoring measures the precise scope and duration of human-authorized AI actions and decisions, and hits a kill switch when it revokes an AI-issued certificate.

But a kill switch, it says, is not governance. Regulations are developing, but cannot keep pace.

“In The Terminator, humanity had to send someone back in time to regain control. Enterprises do not have that luxury. The time to establish control is now – before agentic AI becomes so deeply embedded in critical workflows that intervention means shutting down the business.”

D-ID’s ‘highly expressive avatars’ make AI agents look like real humans

Your bot has a new face: D-ID has announced the launch of V4 Expressive Visual Agents. A release says the product offers “a new generation of ultra-high-fidelity digital humans designed for real-time, LLM-connected conversations, as well as scripted long-form enterprise video content.”

The fresh avatars are built on a new diffusion-based model, and trained on performances captured from real actors to deliver faster generation, low latency (sub-0.5-second) conversational turns, and highly accurate lip sync.

Per the release, V4 Avatars are engineered specifically for low-latency delivery, making them suitable for extended real-time conversations, as well as longer-form content such as training modules, explainers, and multilingual educational videos. The fake people “dynamically align with selected sentiments, ensuring that tone and intent match the underlying message.”

“We have come a long way since our first models that delighted the world by turning still images into talking portraits,” says D-ID CEO Gil Perry. “Today, with V4, we’re setting a new benchmark for avatar fidelity and performance while keeping it fast enough for real-time conversations and consistent, efficient and secure enough for enterprise scale. This advancement in avatar technology positions D-ID as the frontrunner in providing the visual interface layer for the next wave of AI adoption as businesses seek to make interactions more natural and humanlike.”

Darwinium monitors behavior to gauge friction

AI fraud prevention platform Darwinium has launched a new intent-based authentication and orchestration product to secure agentic commerce.

A release says the tool can be deployed natively at the edge across major CDN providers to determine whether a request originates from a verified AI agent, a human user, or malicious automation. It then analyzes real-time behavioral signals and journey context and dynamically applies the appropriate level of friction, permitting low-risk agent interactions such as browsing and product discovery, and challenging risk activity through silent or step-up authentication.

“AI agents are already acting across ecommerce and financial services,” says Alisdair Faulkner, CEO of Darwinium. “The real challenge is distinguishing trusted automation from abuse and applying the right level of trust at the right moment.”

Section2, authID strategic partnership targets financial crime

Biometric identity verification and AI governance provider authID has announced a strategic partnership with financial threat intelligence firm Section2, Inc. A release says the collaboration integrates authID’s biometric identity platform and Mandate governance framework for agentic AI into Section 2’s core solutions, TENet and TRACC, which “generate sensitive intelligence and risk mappings that transform how financial institutions detect and respond to threats.”

The partnership aims to ensure that intelligence delivered into financial institution workflows is cryptographically bound to verified human or agentic identities. “This creates an immutable chain of custody and accountability from intelligence creation through deployment.”

“Section 2 is solving one of the hardest problems in financial crime, turning noise into signal,” says Rhon Daguro, CEO of authID. “By embedding biometric identity verification and our authID Mandate AI governance framework directly into their platform, we ensure every intelligence output is authorized, traceable, and secure. In high-risk financial environments, auditability is not optional, it’s foundational.”

Docusign, Cowork enable users to talk out contracts

Docusign has announced that its Intelligent Agreement Management (IAM) platform is available as part of Anthropic’s Cowork. A release says the integration allows businesses to securely create, review, send and manage agreements from start to finish, through natural language prompts in Cowork.

Potential use cases include using natural language to draft contracts from template, surfacing all expiring customer contracts, requesting reports and initiating new customer onboarding workflows.

“Docusign and Anthropic are raising the bar for agreement actions,” says Allan Thygesen, CEO of Docusign. “What Docusign brings to agentic experiences like Cowork is deep context across all business agreements – the intelligent workflows that know how to act on that context and the trust, security, and scale enterprises expect.”

Article Topics

AI agents  |  authID  |  biometric authentication  |  biometrics  |  D-ID  |  Darwinium  |  digital identity  |  DocuSign  |  identity access management (IAM)  |  Tools for Humanity  |  World  |  ZeroBiometrics