Oxford Research Network Apologizes for AI Screening of Grant Applications

An Oxford-based cybersecurity network apologized for a lack of transparency over AI screening of grant applications and will have humans re-review all rejected proposals.

maisiekooc
Maisie Morrison

AgentLocker Editor

AI News
Oxford Research Network Apologizes for AI Screening of Grant Applications

A research network based at the University of Oxford has apologized for how it used artificial intelligence to screen funding applications. The group said all rejected proposals will now be reviewed again by humans.

The network, called Crane, supports cybersecurity research in the UK. It used what it called an "AI-based triage" to screen out about half of all applicants to one of its funding calls.

According to an email seen by Research Professional News, offers already given to successful applicants are now "suspended."

What Happened With the Funding Call

Crane is backed by a £6.5 million grant from the Engineering and Physical Sciences Research Council, known as EPSRC. The network uses that money to run its own funding calls.

The call in question opened in June and is worth £580,000. It focuses on "securing trust, autonomy and resilience in future cyber ecosystems."

Last month, Crane told unsuccessful applicants that around half of all proposals had been removed through the AI screening process.

After they were rejected, some researchers told Research Professional News they had concerns. These included questions about transparency and how their data was handled.

Crane said the confidentiality of all proposals was protected throughout the process.

Apology and New Review Process

On October 2, Crane sent a new message to rejected applicants. In it, the network apologized for how it handled the process.

"Although this process was undertaken with care, including human sampling and statistical cross-checking, and with appropriate safeguards for confidentiality, we recognise that we were not sufficiently transparent about our use of AI in the triage process," wrote Andrew Martin. "We apologise for this."

Martin is the principal investigator at Crane. He is also a professor of systems security at the University of Oxford.

In the letter, Martin said Crane will re-review every rejected application using human reviewers. He said Crane and EPSRC agreed that the "complexities and sensitivities" of using AI in peer review called for this step.

He added that "all practical steps" will be taken so reviewers do not know how the AI ranked each proposal before.

One version of the notice told an applicant that "offers previously issued are suspended."

"We recognise the importance of transparency and trust in the review process, and we are sorry that our initial communication did not provide sufficient clarity about the approach to be taken," Martin wrote.

"We hope the full human peer review process outlined above provides a clear and fair way forward," he added.

Research Professional News has contacted UK Research and Innovation, which oversees EPSRC, for comment. No response had been reported at the time of writing.

From our research desk
AI Jobs Automation Index
Which jobs are AI tools targeting most? We mapped 3,400+ AI tools to real job functions — with BLS employment & salary data.
Explore the index
maisiekooc

Written by

Maisie is a news writer at Agent Locker, covering the latest developments in artificial intelligence, emerging technology and the companies shaping the future.

Discover AI Agents