The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Agentic AI is safer when several kinds of organizations work together rather than when one vendor writes its own rules. Government-led programs are developing measurement and standards; security coalitions are publishing secure-design practices and open tooling; identity specialists are defining controlled delegation for logins and payments; and evaluation groups are creating reliability benchmarks. The main alliances are NIST’s U.S. AI Safety Institute Consortium (AISIC), NIST’s AI Agent Standards Initiative, the Coalition for Secure AI (CoSAI), the FIDO Alliance, the MLCommons ARES collaboration, and Google’s Action Blueprint.
They are complementary, not interchangeable. AISIC and the new NIST initiative concentrate on measurement, risk and standards leadership. CoSAI focuses on engineering and governance patterns, FIDO on authentication and delegated commerce, MLCommons on repeatable reliability evaluation, and Google’s blueprint on coordinated public- and private-sector implementation. An organization deploying agents should use their outputs as layers in one assurance program, while treating proposals and emerging guidance as work in progress rather than certification.
Which alliances are working to make agentic AI safer?
Six initiatives currently cover the most important parts of the safety stack, from measuring capabilities to authorizing a purchase. Their mandates and maturity differ, so the useful question is not which single group “owns” agent safety, but which layer each group can strengthen.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
| Alliance or initiative | Governance model | Primary technical scope | Typical outputs | How organizations participate |
|---|---|---|---|---|
| NIST U.S. AI Safety Institute Consortium (AISIC) | U.S. government-led consortium | Red-teaming, capability evaluation, risk management, safety and security, watermarking and AI-safety measurement | Measurement foundations, evaluation work and risk guidance | Membership spanning companies, researchers, government, academia and civil society |
| NIST AI Agent Standards Initiative | Government standards initiative with international and community tracks | Agent standards, open-source protocols, agent security and identity | Standards input, protocol development and future guidance | Requests for information, convenings and listening sessions |
| Coalition for Secure AI (CoSAI) | OASIS Open coalition | AI software supply-chain security, defender preparation, security-risk governance and secure agent design | Secure-design patterns, governance material and open technical work | Free technical participation for developers; sponsorship is available |
| FIDO Alliance | Open identity and authentication standards alliance | Verifiable instructions, agent authentication and trusted delegation for commerce | Extensions, profiles, architectural frameworks and implementation practices | Participation through its technical working groups |
| MLCommons ARES collaboration | Open, multi-stakeholder evaluation collaboration | Agent reliability across correctness, safety, security and control | Design principles and business-relevant benchmarks | AI labs, deployers, testing providers, universities and policy organizations contribute |
| Google’s Action Blueprint | Industry-led proposal | Coordinated standards, pilots, information sharing, infrastructure upgrades and resilience | Proposed protocols, secure sandboxes, procurement and incident-readiness actions | Work through initiatives such as CoSAI, ISO/IEC and NIST |
The boundaries overlap deliberately. A benchmark can expose a security weakness; a protocol standard can make the fix interoperable; and procurement requirements can turn both into a deployment gate.
#1 Best Overall
- Read Before You Buy — No Video Output: These adapters support charging and USB 2.0 data transfer, but cannot transmit video signals. Except for standard USB webcams (which use USB data only), they are not compatible with HDMI/DisplayPort cables, video-capable USB-C hubs, or docking stations with video output.
- Convert USB-A Ports to USB-C: Designed to connect USB-C earphones, cables, flash drives, card readers, and other USB-C accessories to standard USB-A ports. Plug-and-play with no drivers or software required.
- Aluminum Alloy Housing: Built with a sturdy aluminum alloy shell that aids in heat dissipation and protects against daily wear and scratches. Designed to maintain a stable and secure connection.
- Compact & Travel-Friendly: The ultra-compact design allows the adapter to stay plugged into your device without blocking adjacent ports or adding bulk, reducing wear and tear on your original USB ports.
- 12-Month Warranty: Backed by a 12-month manufacturer warranty for peace of mind. Designed to meet strict quality control standards for reliable everyday performance.
What does NIST’s U.S. AI Safety Institute Consortium contribute?
NIST announced AISIC on February 8, 2024. It brings together more than 200 member companies and organizations, including AI creators and users, academic and government researchers, and civil-society groups. Its breadth matters because an agent’s risk is affected by its model, tools, data, hosting environment and real-world users—not just by the model developer.
AISIC’s work covers red-teaming, capability evaluations, risk management, safety and security, watermarking and the foundations of AI-safety measurement. These activities help organizations turn vague claims such as “the agent is safe” into testable questions: What can it do under adversarial prompting? Which tools can it reach? How reliably does it follow a permitted instruction? Can an evaluator reproduce the result?
NIST Director Laurie E. Locascio described the measurement problem directly: “AI is moving the world into very new territory. And like every new technology, or every new application of technology, we need to know how to measure its capabilities, its limitations, its impacts.” For an enterprise, AISIC is therefore most relevant when establishing evaluation methods and risk terminology before selecting a specific control.
What is the NIST AI Agent Standards Initiative?
Announced on February 17, 2026, this initiative addresses agents that can take autonomous actions. NIST identifies three pillars: industry-led agent standards and international standards leadership; community-led development and maintenance of open-source protocols; and research on agent security and identity.
Agents create problems that ordinary chat systems do not. They interact with external services, invoke tools, move data between systems and may act over time without a person reviewing every intermediate step. NIST says those properties create reliability and interoperability challenges. A payroll agent, for example, needs a common way to represent an approved instruction, discover a permitted service and prove which identity authorized the action.
The initiative is using convenings, requests for information and listening sessions to shape guidance. That process means its direction is influential but still developing. Organizations should contribute concrete deployment scenarios and failure data, then design systems that can adopt changing standards instead of assuming a final specification already exists.
How does CoSAI address secure agent development?
The Coalition for Secure AI is an OASIS Open coalition of AI and security experts. Its workstreams cover software supply-chain security for AI systems, preparing defenders, AI-security risk governance and secure design patterns for agentic systems.
Windows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallCrashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteRank #2
- 5-in-1 USB-C Hub: Experience comprehensive connectivity featuring a Power Delivery input, two USB-A 2.0 ports, a USB-A 3.0 port, and an HDMI port. (Note: The USB-C power delivery input port is only for connecting an external wall charger to power your laptop and cannot power peripheral devices.)
- 90W Pass-Through Charging: Achieve optimal charging with 90W pass-through power to your laptop, supported by a total input of 100W, with the hub reserving 10W for operational efficiency. (Note: Wall charger not included.)
- Quick Data Transfers: Accelerate your productivity with rapid data transfers using a high-speed 5Gbps USB 3.0 port and two 480Mbps USB 2.0 ports.
- 4K HDMI Display: Enhance your visual experience with a hub capable of delivering 4K resolution at 30Hz in both mirror and extend modes. Please note that this hub is compatible with MacBook (macOS 12 and newer), Windows 10 and 11, ChromeOS, and laptops equipped with DP Alt Mode and Power Delivery. Note: This device is not compatible with Linux.
- What You Get: Anker USB-C Hub (5-in-1, 4K HDMI), welcome guide, 18-month warranty, and our friendly customer service.
Software supply-chain security
Agent applications combine models, prompts, retrieval components, plugins, tool connectors, container images and ordinary software dependencies. A compromise in any of those layers can alter an agent’s behavior. CoSAI’s supply-chain focus is useful for establishing provenance, review and update controls across the complete application rather than auditing only the model.
Defender preparation and governance
Preparing defenders means giving security teams practical ways to monitor, investigate and contain agent behavior. Governance work connects technical controls to decisions about who may approve tools, what evidence must be retained and when an agent must be taken offline.
Secure-by-design patterns
Secure design patterns are reusable architecture choices, such as limiting an agent to least-privilege tools, separating planning from execution and requiring explicit approval for high-impact actions. CoSAI says technical participation is free and open to developers, with sponsorship available. That makes it a practical venue for engineers who want to contribute implementation experience rather than only comment on policy.
How is the FIDO Alliance solving agent authentication and delegated payments?
FIDO is developing open, interoperable standards for agentic authentication and commerce. Its approach starts with a basic rule: an agent should be able to act on a user’s instructions without receiving the user’s raw password, private key or payment credential.
Verifiable user instructions
An instruction should carry evidence of what the user approved, for which service, under what limits and for how long. The receiving service needs to distinguish a current, authentic instruction from a replayed message or an instruction silently changed by an intermediary.
Authentication without exposing raw credentials
Rather than handing credentials to an agent, a user can authenticate through a mechanism that lets the service verify authorization while keeping the underlying secret protected. This reduces the damage if an agent’s context, log or tool output is compromised.
Trusted delegation for commerce
FIDO’s Agentic Authentication and Payments technical working groups are developing extensions, profiles, architectural frameworks and practices for delegated actions that are controlled, auditable and revocable. Those properties are essential for purchases: a user should be able to set a merchant, spending limit, product category or expiration, inspect what happened and cancel future authority.
Rank #3
- Sleek 7-in-1 USB-C Hub: Features an HDMI port, two USB-A 3.0 ports, and a USB-C data port, each providing 5Gbps transfer speeds. It also includes a USB-C PD input port for charging up to 100W and dual SD and TF card slots, all in a compact design.
- Flawless 4K@60Hz Video with HDMI: Delivers exceptional clarity and smoothness with its 4K@60Hz HDMI port, making it ideal for high-definition presentations and entertainment. (Note: Only the HDMI port supports video projection; the USB-C port is for data transfer only.)
- Double Up on Efficiency: The two USB-A 3.0 ports and a USB-C port support a fast 5Gbps data rate, significantly boosting your transfer speeds and improving productivity.
- Fast and Reliable 85W Charging: Offers high-capacity, speedy charging for laptops up to 85W, so you spend less time tethered to an outlet and more time being productive.
- What You Get: Anker USB-C Hub (7-in-1), welcome guide, 18-month warranty, and our friendly customer service.
These efforts are standards work, not a promise that every merchant or agent already supports a common payment flow. Deployers should ask which profiles a service implements and how it records consent, limits and revocation.
What does MLCommons ARES measure in an AI agent?
MLCommons announced its open agent-reliability evaluation collaboration on June 27, 2025. Thirty-four organizations attended its Agentic Reliability Evaluation Summit, representing AI labs, enterprise deployers, testing providers, universities and policy organizations.
The ARES framework is organized around four dimensions:
- Correctness: whether the agent reaches the intended result and uses tools appropriately.
- Safety: whether it avoids harmful actions and respects policy constraints.
- Security: whether it resists attacks, protects data and limits unauthorized access.
- Control: whether people can constrain, observe, interrupt and recover its behavior.
Its planned outputs include design principles and business-relevant benchmarks. That emphasis helps teams compare an agent against tasks they actually run—such as reconciling invoices or opening support tickets—rather than relying on a generic language benchmark. A score is still only evidence for a defined scenario; it is not a universal safety certification.
What is Google’s Action Blueprint, and how authoritative is it?
Google’s Action Blueprint, proposed on October 6, 2026, calls for coordinated standards and protocols; public-sector pilots and secure sandboxes; information sharing; infrastructure and code upgrades; resilience and incident-response readiness; and secure-by-design procurement.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →Scan for outdated or missing drivers - takes under a minuteDriver Scan →The proposal is notable for connecting technical standards to deployment mechanisms. Pilots can expose failures in a controlled environment, information sharing can shorten the time between an incident and a defensive update, and procurement language can require suppliers to demonstrate controls before an agent reaches production.
It remains a corporate proposal, not a settled international standard. Google says it will work through multi-stakeholder initiatives including CoSAI, ISO/IEC and NIST, but each participating body retains its own process and authority. Treat the blueprint as a coordination agenda and a source of procurement ideas, not as a certification scheme.
Rank #4
- Dual Converters, Infinite Potential：Includes 2× USB C male to USB A female adapters and 2× USB A male to USB C female adapters. Perfect for a wide range of uses—tablets with Bluetooth keyboards, expand USB ports on macbook, and more. Two different converters for all your daily needs
- Next-Level 10Gbps & 3A Charging: No more slow 480Mbps, this usb to usb c adapter has a transfer speed of up to 10Gbps, allowing you to do more transferring in less time. This usb adapter fits both USB A and USB C charger, supporting up to 3A fast charging
- Upgraded Exquisite Craftsmanship: With an aluminum alloy housing and metal connector, the usbc to usb adapter is extremely durable and sturdy. Rigorously tested to withstand more than 10,000 times of plugging and unplugging, ensuring long-lasting performance
- Broad Compatible: The usb c to usb adapter widely supports all USB C/ USB A devices like laptops, tablets, cellphones, car chargers, and phone chargers. Such as compatible with MacBook Pro/Air 2023/2022, Thunderbolt 4/3 Devices,Apple MagSafe Watch 9/8/7/SE/Ultra, iPad Pro 2022/2021, Samsung Galaxy S23/S20/S10, and iPhone 17/16/15 Pro. Plug and play
- Please Note: To reach 10Gbps speed, keep the cable under 3.3 ft. For USB A Male to USB C adapters, try flipping the USB C connector. USB C Male to USB A adapters support bidirectional 10Gbps transfer within 3.3 ft
Who sets standards for agentic AI security?
No single alliance sets all agentic-AI security standards. NIST provides government convening and measurement leadership; international standards bodies and industry groups develop interoperable specifications; open coalitions publish engineering patterns; identity alliances define delegated authentication; and evaluation consortia make reliability claims testable.
Compare an initiative on four axes before relying on its output:
- Governance model: determine whether it is a government consortium, standards alliance, open coalition or industry-led proposal, and therefore what authority its documents carry.
- Technical scope: identify whether the work addresses measurement, protocols, identity, secure development or reliability evaluation.
- Participation route: check whether the contribution mechanism is a public consultation, formal membership, technical working group or sponsorship.
- Output type: distinguish guidance, a normative standard, open-source code, a benchmark or a procurement recommendation.
This distinction prevents a common mistake: treating a useful design pattern or draft benchmark as if it were a mandatory rule.
How do I evaluate an AI agent before deployment?
Use the alliances as complementary inputs to a staged assessment. The following sequence turns their themes into a practical deployment gate.
- Define the agent’s authority. List every system it can read, write, call or purchase from, the data it can expose, and the actions that require a human approval.
- Set measurable tasks. Write business scenarios with an expected result, allowed tools, time limit and unacceptable outcomes. These become your correctness tests and make later benchmark results meaningful.
- Run adversarial evaluation. Red-team prompts, retrieved documents, tool responses and identity boundaries. Test prompt injection, data exfiltration, unauthorized tool use and attempts to bypass approval.
- Score the four reliability dimensions. Record separate evidence for correctness, safety, security and control instead of collapsing them into one pass/fail number.
- Review the software supply chain. Inventory model versions, libraries, connectors, plugins, prompts and deployment images. Require provenance, code review and a rollback path for updates.
- Design least-privilege execution. Give each tool the smallest permissions and narrowest data scope needed. Separate planning from execution where possible, and require confirmation for irreversible or high-impact actions.
- Bind actions to identity and consent. Use verifiable instructions and delegated authentication so the agent does not receive raw credentials. Record the principal, approved scope, expiration and transaction details.
- Pilot in a sandbox. Start with synthetic or minimized data and simulated external systems. Observe behavior, latency, tool errors and escalation frequency before allowing production effects.
- Prepare response and recovery. Define who can pause the agent, revoke delegated authority, rotate affected credentials, restore data and notify stakeholders. Keep logs that support a reconstruction of each decision and tool call.
- Make procurement enforceable. Put required tests, evidence, update notices, incident reporting and revocation support into contracts. Re-test after model, tool or policy changes.
This process does not eliminate risk. It gives a deployment team observable evidence and a way to stop or correct an agent when its environment changes.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How can an agent authenticate or make a purchase on my behalf?
Use delegated authority rather than giving the agent a standing password or unrestricted payment token. The user first authenticates and specifies a bounded instruction; the service verifies that instruction; the agent performs only the permitted action; and the authority can be audited and revoked.
Recommended Free Tools
For authentication
- Define the service, action, data scope and expiration.
- Require a user-verifiable authorization record instead of accepting an instruction copied from untrusted agent context.
- Keep raw credentials outside the agent’s working memory, logs and tool responses.
- Require step-up confirmation when the action exceeds the original scope.
For commerce
- Set merchant, category, amount and frequency limits before delegation.
- Show the user the final transaction details at the point required by the policy.
- Record the authorization, the agent identity, the merchant response and the result.
- Provide immediate revocation and a way to dispute or reverse an unauthorized action.
FIDO’s working groups are developing the standards needed to make these controls interoperable. Until services support compatible profiles, organizations must verify the exact mechanism each agent and merchant implements.
Best Value
- 5-in-1 Connectivity: Equipped with a 4K HDMI port, a 5 Gbps USB-C data port, two 5 Gbps USB-A ports, and a USB C 100W PD-IN port. Note: The USB C 100W PD-IN port supports only charging and does not support data transfer devices such as headphones or speakers.
- Powerful Pass-Through Charging: Supports up to 85W pass-through charging so you can power up your laptop while you use the hub. Note: Pass-through charging requires a charger (not included). Note: To achieve full power for iPad, we recommend using a 45W wall charger.
- Transfer Files in Seconds: Move files to and from your laptop at speeds of up to 5 Gbps via the USB-C and USB-A data ports. Note: The USB C 5Gbps Data port does not support video output.
- HD Display: Connect to the HDMI port to stream or mirror content to an external monitor in resolutions of up to 4K@30Hz. Note: The USB-C ports do not support video output.
- What You Get: Anker 332 USB-C Hub (5-in-1), welcome guide, our worry-free 18-month warranty, and friendly customer service.
Which alliance should an organization engage first?
Choose according to the problem you need to solve, then combine the result with the other layers:
- Building an evaluation program: start with AISIC’s measurement and red-teaming orientation, then use MLCommons ARES dimensions to structure repeatable business tests.
- Engineering an agent platform: use CoSAI’s secure-design and supply-chain work, while tracking NIST’s protocol and identity direction for interoperability.
- Delegating logins or payments: follow FIDO’s authentication and commerce work and verify support for its emerging profiles.
- Shaping public policy or international standards: participate in NIST’s convenings, information requests and listening sessions, and distinguish draft input from adopted standards.
- Writing enterprise procurement requirements: use the Action Blueprint’s pilot, information-sharing, resilience and secure-procurement ideas, then turn them into measurable contract obligations.
Engagement is most valuable when participants bring concrete failure cases, test data and implementation constraints. Abstract endorsements do less to improve safety than reproducible evidence.
What these alliances cannot guarantee
Membership, attendance or alignment with an alliance does not prove that a particular agent is safe. None of the initiatives described here supplies a universal product certification, and a benchmark result applies only to its defined tasks, data and test conditions.
Standards can also lag fast-changing agent capabilities. A protocol may be interoperable yet incorrectly configured; a secure design pattern may fail when a connected tool behaves unexpectedly; and a human approval step may become meaningless if reviewers cannot understand the requested action. Continuous monitoring, scoped permissions, independent testing and a tested shutdown path remain necessary after launch.
The practical outcome is a layered assurance program: measure capabilities, engineer secure components, authenticate delegated authority, evaluate reliability in realistic tasks and make suppliers accountable for updates and incidents. The alliances make those layers more consistent; they do not remove the need for local risk decisions.
Frequently Asked Questions
Are any of these agent-safety alliances regulators?
The initiatives described are consortiums, standards groups, collaborations or a corporate proposal rather than a single regulatory authority. Their documents can influence regulation and procurement, but legal duties still come from the jurisdictions and sectors in which an agent operates.
Can a small developer participate without joining a large consortium?
Yes. CoSAI states that technical participation is free and open to developers. NIST’s agent initiative also gathers input through public requests for information, convenings and listening sessions, while other groups provide working-group or collaboration routes.
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Fix the driver behind crashes, sound loss and screen glitches3Repair Windows errors before they cause bigger problemsWhat evidence should an AI-agent vendor provide during due diligence?
Ask for task-level evaluation results, red-team findings, a component and dependency inventory, tool-permission documentation, delegated-authentication details, logging and revocation procedures, update notices, and incident-response contacts. Require the evidence to identify the tested version and operating conditions.
How often should an agent be re-evaluated?
Re-test whenever the model, prompt policy, tool connector, data source, identity flow or execution environment changes, and after a material incident. A fixed calendar review can supplement those triggers but should not replace change-based testing.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
