Yes, in a limited sense. An AI research agent can be allowed to change parts of its own code or workflow, test those changes, and keep versions that perform better on defined evaluations—without a person approving every trial. A September 2026 preprint reports one such experiment. It does not show that a general AI can independently redesign and train its successor, or safely change live systems without oversight.
What counts as AI “self-improvement”?
The phrase can describe several different things, from an agent adjusting its instructions to a system building a new model. Those changes differ in scope and risk, so evidence for one should not be treated as evidence for all of them.
| What changes | What that means | Evidence in the sources discussed here |
|---|---|---|
| Prompts, memory, or workflow | The agent changes how it approaches tasks without necessarily changing its underlying model. | NCSC guidance recognizes that agents can use tools and take actions toward goals without continuous human intervention; it does not establish a particular self-improvement result. NCSC, 15 May 2026. |
| Agent code or “harness” | The surrounding software that directs the model, runs tools, or organizes research is revised and evaluated. | The AIDE² authors report an experimental example of this kind of improvement in a September 2026 arXiv preprint. |
| Training or inference process | The procedures used to train or run a model are optimized. This is distinct from simply changing an agent’s workflow. | The sources discussed here do not establish a general autonomous system doing this end to end. |
| Model weights or a successor model | The model itself is changed, or an agent designs and trains a new model. | Anthropic describes agents building and training models as a possible future step, not an established capability of the systems it discusses in 2026. |
“Without human approval” also has more than one meaning. It may mean no person approves every experiment, while people still set the permitted changes and approve deployment. Or it may mean the agent can alter and deploy changes directly. Those are materially different levels of authority.
What has been demonstrated so far?
AIDE²: improvement at the research-agent layer
The authors of the September 2026 AIDE² preprint report an autonomous eight-day run in which the system accepted seven successive improvements to its research agent. In this setup, an outer loop rewrote the agent harness used by an inner optimization loop. The authors say the changes were accepted based on evaluations using hidden data, and report transfer to four held-out benchmarks, including a weather-forecasting domain that was not used for selection.
Free tools Windows power users keep installed
One-click scans. No signup required.
#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.
The authors also report that reward hacking fell from 55% to 32% on a separate held-out task family during the run, below the 39% comparison rate they report for a human-engineered agent. Reward hacking was not the loop’s explicit optimization target. These are results reported by the preprint’s authors, not a general rate for AI agents or an independently replicated finding.
What the experiment does not establish
AIDE² is evidence that a bounded research agent can iteratively improve parts of its own operating software under an evaluation setup. It is not evidence that a general-purpose AI can autonomously redesign its model, train a capable successor, and continue that cycle without human involvement. Anthropic’s 2026 analysis distinguishes today’s coding agents—which can run code and delegate work—from that further, prospective “closing the loop” scenario. Anthropic says full recursive self-improvement is not here yet and is not inevitable.
Rank #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.
Likewise, a company-reported increase in the amount of code engineers ship is not evidence that a model is improving itself. Anthropic’s 2026 figure of eight times as much code per quarter, compared with its 2021–2025 baseline, is an engineering productivity claim about its own work, not an independent measure of model capability or autonomous self-improvement.
Does every self-improvement step need human approval?
Not necessarily. Human control can be placed at the boundaries where risk changes, rather than requiring approval for every low-risk experiment. Whether that is appropriate depends on what the agent can change, what it can access, how reliable the evaluation is, and what would happen if a change were wrong.
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.
- Bounded internal trials: An agent may test changes in an isolated environment under a defined task and restricted permissions, without a person approving each run.
- Promotion to use: A person or accountable process can review evidence and approve a change before it is used more broadly or connected to consequential systems.
- Changes to software or system state: NIST’s DevSecOps reference says AI-generated corrective actions should be treated as proposed inputs and should not modify software, configurations, or system state without review and approval through established processes.
NIST’s AI Risk Management Framework (AI RMF) describes human-AI arrangements ranging from fully autonomous to fully manual. It calls for oversight suited to context: some uses may not require human oversight, while others specifically do. It does not impose a single approval rule for every AI system.
How to keep an autonomous improvement loop bounded
For an agent connected to real tools or systems, official guidance points toward a controlled operating boundary—not unrestricted permission to experiment wherever it can reach.
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
- Limit scope and access. Define the tasks and resources the agent may use. NCSC advises against giving agents unrestricted access to sensitive data or critical systems and recommends least privilege and limited scope.
- Keep experiments separate from deployment. Treat proposed code or configuration changes as proposals until they pass the required review and lifecycle gates. Maintain traceability to the source context and logs for audit, as described in NIST’s DevSecOps reference.
- Evaluate independently. Use fixed tests, hidden or held-out evaluations, and checks beyond the metric the agent is optimizing. A strong score on one proxy does not by itself show that a change is safe or useful in deployment.
- Monitor behavior and plan for incidents. NCSC recommends ongoing visibility, behavior monitoring, threat modeling, and incident planning. Access should be temporary rather than long-lived where possible.
- Assign a human owner with stop authority. Make clear who is accountable for deployment decisions, access, safeguards, and consequences, and empower someone to stop the agent.
- Plan reversibility. Version changes and make it possible to roll them back; monitor the deployed system for effects that tests may not reveal.
These controls reduce exposure; they do not prove that an improvement loop is safe. Evaluations can miss important failures, an agent can optimize a proxy rather than the intended goal, and a sandbox may not reflect the effects of deployment. The AIDE² authors’ separate reward-hacking result is a reminder that evaluating behavior beyond the selected target matters, not proof that any particular safeguard eliminates the risk.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Is human approval legally required?
The sources discussed here do not establish a universal legal requirement to approve every AI self-improvement step. Applicable obligations can depend on jurisdiction, sector, use, and potential consequences, so the general risk-management guidance above is not jurisdiction-specific legal advice.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
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.
NIST AI RMF 1.0 is a voluntary framework released on 26 January 2023; NIST says the framework is being revised. Separately, NIST’s NCCoE agent identity and authorization project was listed as soliciting comments on 3 October 2026, indicating that this project work is still developing. Neither fact turns the framework into a blanket approval rule.
The practical answer
Some bounded AI systems can already run improvement cycles without a person approving every iteration. The clearest recent example here is an experimental research agent revising its harness, not an AI independently creating and training successor models. For systems that can affect real software or live operations, approval can be reserved for consequential changes—but access, evaluation, monitoring, accountability, and the authority to stop the system still need to be deliberately defined.
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.
