(Press-News.org) Investigation of 30 top AI agents finds just four have published formal safety and evaluation documents relating to the actual bots.
The new wave of AI web browser agents, many designed to mimic human browsing, have the highest rates of missing safety information.
Many of us now use AI chatbots to plan meals and write emails, AI-enhanced web browsers to book travel and buy tickets, and workplace AI to generate invoices and performance reports.
However, a new study of the “AI agent ecosystem” suggests that as these AI bots rapidly become part of everyday life, basic safety disclosure is “dangerously lagging”.
A research team led by the University of Cambridge has found that AI developers share plenty of data on what these agents can do, while withholding evidence of the safety practices needed to assess any risks posed by AI.
The AI Agent Index, a project that includes researchers from MIT, Stanford and the Hebrew University of Jerusalem, investigated the abilities, transparency and safety of thirty “state of the art” AI agents, based on public information and correspondence with developers.
The latest update of the Index is led by Leon Staufer, a researcher studying for an MPhil at Cambridge’s Leverhulme Centre for the Future of Intelligence. It looked at available data for a range of leading chat, browser and workflow AI bots built mainly in the US and China.
The team found a “significant transparency gap”. Developers of just four AI bots in the Index publish agent-specific “system cards”: formal safety and evaluation documents that cover everything from autonomy levels and behaviour to real-world risk analyses.*
Additionally, 25 out of 30 AI agents in the Index do not disclose internal safety results, while 23 out of 30 agents provide no data from third-party testing, despite these being the empirical evidence needed to rigorously assess risk.
Known security incidents or concerns have only been published for five AI agents, while “prompt injection vulnerabilities” – when malicious instructions manipulate the agent into ignoring safeguards – are documented for two of those agents.
Of the five Chinese AI agents analysed for the Index, only one had published any safety frameworks or compliance standards of any kind.
“Many developers tick the AI safety box by focusing on the large language model underneath, while providing little or no disclosure about the safety of the agents built on top,” said Cambridge University’s Leon Staufer, lead author of the Index update.
“Behaviours that are critical to AI safety emerge from the planning, tools, memory, and policies of the agent itself, not just the underlying model, and very few developers share these evaluations.”
In fact, the researchers identify 13 AI agents that exhibit “frontier levels” of autonomy, yet only four of these disclose any safety evaluations of the bot itself.
“Developers publish broad, top-level safety and ethics frameworks that sound reassuring, but are publishing limited empirical evidence needed to actually understand the risks,” Staufer said.
“Developers are much more forthcoming about the capabilities of their AI agent. This transparency asymmetry suggests a weaker form of safety washing.”
The latest annual update provides verified information across 1,350 fields for the thirty prominent AI bots, as available up to the last day of 2025.
Criteria for featuring in the Index included public availability and ease of use, and developers with a market valuation of over US$1 billion. Some 80% of the Index bots were released or had major updates in the last two years.
The Index update shows that – outside of Chinese AI bots – almost all agents depend on a few foundation models (GPT, Claude, Gemini), a significant concentration of platform power behind the AI revolution, as well as potential systemic choke points.
“This shared dependency creates potential single points of failure,” said Staufer. “A pricing change, service outage, or safety regression in one model could cascade across hundreds of AI agents. It also creates opportunities for safety evaluations and monitoring.”
Many of the least transparent agents are AI-enhanced web browsers designed to carry out tasks on the open web on a user’s behalf: clicking, scrolling, and filling in forms for tasks ranging from buying limited-release tickets to monitoring eBay bids.
Browser agents have the highest rate of missing safety information: 64% of safety-related fields unreported. They also operate at the highest levels of autonomy.**
This is closely followed by enterprise agents, business management AI aimed at reliably automating work tasks, with 63% of safety-related fields missing. Chat agents are missing 43% of safety-related fields in the Index.***
Staufer points out that there are no established standards for how AI agents should behave on the web. Most agents do not disclose their AI nature to end users or third parties by default.**** Only three agents support watermarking of generated media to identify it as from AI.
At least six AI agents in the Index explicitly use types of code and IP addresses designed to mimic human browsing behaviour and bypass anti-bot protections.
“Website operators can no longer distinguish between a human visitor, a legitimate agent, and a bot scraping content,” said Staufer. “This has significant implications for everything from online shopping and form-filling to booking services and content scraping.”
The update includes a case study on Perplexity Comet: one of the most autonomous browser-based AI agents in the Index, as well as one of the most high-risk and least transparent.
Comet is marketed on its ability to “work just like a human assistant”. Amazon has already threatened legal action over Comet not identifying itself as an AI agent when interacting with its services.
“Without proper safety disclosures, vulnerabilities may only come to light when they are exploited,” said Staufer.
“For example, browser agents can act directly in the real world by making purchases, filling in forms, or accessing accounts. This means that the consequences of a security flaw can be immediate and far-reaching.”
Staufer points out that last year, security researchers discovered that malicious content on a webpage could hijack a browser agent into executing commands, while other attacks were able to extract users' private data from connected services.
Added Staufer: “The latest AI Agent Index reveals the widening gap between the pace of deployment and the pace of safety evaluation. Most developers share little information about safety, evaluations, and societal impacts.”
“AI agents are getting more autonomous and more capable of acting in the real world, but the transparency and governance frameworks needed to manage that shift are dangerously lagging.”
Notes:
* In the Index, only ChatGPT Agent, OpenAI Codex, Claude Code, and Gemini 2.5 Computer Use provide agent-specific system cards, though some Chinese agents have research papers focused on computer-use capabilities.
** Many browser agents operate at levels 4-5 of the “five levels of escalating agent autonomy”. (Level 4 = user as an approver; level 5 = user as an observer).
*** Safety-related fields used by researchers include: guardrails; “sandboxing” (running an AI system inside a controlled environment); risk evaluation; safety evaluations; third-party testing; benchmarks; “bug bounty” (crowdsourced security audit); record of known incidents.
**** 21 of the 30 agents in the Index have no documented default disclosure behaviour.
Full list of collaborating institutions on the AI Agent Index as follows: University of Cambridge (UK), University of Washington (US), Harvard Law School (US), Stanford University (US), Concordia AI (China), University of Pennsylvania (US), Massachusetts Institute of Technology (US), Hebrew University of Jerusalem (Israel).
END
Most AI bots lack basic safety disclosures, study finds
2026-02-20
ELSE PRESS RELEASES FROM THIS DATE:
How competitive gaming on discord fosters social connections
2026-02-20
Human beings are social animals; they need places to relax, connect with others, and feel a sense of belonging beyond the demands of home and work. Traditionally, these ‘third places’ are thought to be limited to cafes, clubs, gardens, and other neighborhood community spots. However, with an increase in digitally shaped life schedules, a new question arises: Do online spaces offer the same social values as offline ones?
“Video games and social media are often considered to be isolating or overly ...
CU Anschutz School of Medicine receives best ranking in NIH funding in 20 years
2026-02-20
The University of Colorado Anschutz School of Medicine continues to boost its standing in the Blue Ridge Institute for Medical Research (BRIMR) rankings, moving to No. 8 among public medical schools in the nation and to No. 21 among all medical schools, with half of its 18 clinical Departments landing in the top 15.
Annual Blue Ridge rankings are determined by total funding granted from the National Institutes of Health (NIH) to researchers at U.S. medical schools and their departments.
This year, BRIMR ranked the Department of Pediatrics No. 1 with more than $63 million in funding from the NIH, the Department ...
Mayo Clinic opens patient information office in Cayman Islands
2026-02-20
GEORGE TOWN, Cayman Islands — Mayo Clinic is opening a representative office in the Cayman Islands to help patients, their families, health insurers and others interested in connecting with Mayo Clinic. The office on Grand Cayman Island is Mayo's first in the Caymans and third in the Caribbean.
The Mayo Clinic Representative Office staff will help patients in the Cayman Islands and surrounding area make appointments at Mayo Clinic in Rochester, Minnesota; Phoenix and Scottsdale, Arizona; ...
Phonon lasers unlock ultrabroadband acoustic frequency combs
2026-02-20
Acoustic frequency combs organize sound or mechanical vibrations into a series of evenly spaced frequencies, much like the teeth on a comb. They are the acoustic counterparts of optical frequency combs, which consist of equally spaced spectral lines and act as extraordinarily precise rulers for measuring light. While optical frequency combs have revolutionized fields such as precision metrology, spectroscopy, and astronomy, acoustic frequency combs utilize sound waves, which interact with materials in fundamentally different ways ...
Babies with an increased likelihood of autism may struggle to settle into deep, restorative sleep, according to a new study from the University of East Anglia.
2026-02-20
When “quiet” isn’t quiet enough: how shallow sleep in infants may relate to autism
Peer reviewed – experimental study - humans
Babies with an increased likelihood of autism may struggle to settle into deep, restorative sleep, according to a new study from the University of East Anglia.
Researchers studied the link between sleep and sensory sensitivity, which is common in neurodivergent infants.
They found that when babies with this trait napped in a noisy environment, their deep sleep was considerably disrupted.
But even ...
National Reactor Innovation Center opens Molten Salt Thermophysical Examination Capability at INL
2026-02-19
(IDAHO FALLS, Idaho) — The National Reactor Innovation Center’s Molten Salt Thermophysical Examination Capability is set to begin operation in March, marking a pivotal step toward advancing reactor and fuel cycle technologies.
MSTEC, located at the Idaho National Laboratory, is a state-of-the-art, shielded argon glove box for irradiated and nonirradiated actinide materials, specifically high-temperature liquids such as fuel salts. This is one of NRIC’s multiple testing capabilities advancing nuclear energy in the United States.
NRIC officially unveiled the capability today during a small ceremony. It was attended ...
International Progressive MS Alliance awards €6.9 million to three studies researching therapies to address common symptoms of progressive MS
2026-02-19
The International Progressive MS Alliance has awarded €6.9 million to three global studies aimed at finding solutions for the most common symptoms experienced by people living with progressive MS, including fatigue, cognitive impairment, pain and mobility.
The Well-Being Phase II Efficacy Awards are part of the Alliance’s overall Well-being Research Pipeline, a large, multi-stage initiative to design, test and implement innovative approaches to solve some of the most difficult aspects of living with progressive MS.
The three projects include:
Comparison of Self-Guided, Coached and Therapist-Delivered Pain Self-Management ...
Can your soil’s color predict its health?
2026-02-19
Determining the health of agricultural soil has traditionally been a slow, messy, and expensive process involving hazardous chemicals. But what if the answer was as simple as the shade of the dirt itself? A groundbreaking study published in Carbon Research reveals that analyzing soil color indices is not only a scientifically sound way to predict Soil Organic Matter (SOM) but also a massive financial win for farmers and laboratories. Led by Dr. Yassine Bouslihim from the National Institute of Agricultural Research (INRA) in Morocco, the study explores the potential of "colorimetric" soil testing in semi-arid agricultural regions. By shifting from traditional chemical-heavy ...
Biochar nanomaterials could transform medicine, energy, and climate solutions
2026-02-19
A new scientific review highlights how emerging biochar-based nanomaterials could play a powerful role in tackling global challenges ranging from climate change to healthcare innovation. Researchers report that nanobiochar and biochar nanocomposites, advanced forms of carbon-rich materials derived from biomass, are showing promise across fields including renewable energy storage, sustainable construction, agriculture, and even medicine.
Biochar itself is produced by heating plant-based waste such as crop residues or forestry byproducts in low-oxygen conditions. When engineered at the nanoscale or combined ...
Turning waste into power: scientists convert discarded phone batteries and industrial lignin into high-performance sodium battery materials
2026-02-19
Researchers have developed a new method to transform two major waste streams, discarded mobile phone batteries and industrial lignin, into a promising material for next-generation sodium-ion batteries. The study demonstrates how waste recycling can simultaneously reduce environmental pollution and support the transition to sustainable energy storage technologies.
Mobile phone batteries are replaced frequently, creating large quantities of electronic waste that contain valuable metals but also pose environmental risks if improperly discarded. At the same time, lignin, a natural polymer generated ...