AI-assisted firms disclose 120% more flaws, fixes keep pace
A Project Glasswing preprint finds companies with AI access reported far more security flaws without any detected slowdown in repairs, though big questions remain.
Advertising disclosure: we may earn a commission when you join an operator via links on this page, at no cost to you. This never dictates our ratings. How we make money
Key takeaways
- A Project Glasswing preprint links AI access to a 120% jump in flaw disclosures.
- Measured fix times showed no detected slowdown.
- Hidden backlogs of unfixed bugs aren't ruled out.
- The study doesn't show AI actually prevents attacks.
Companies with AI access disclosed about 120% more security flaws, and the study behind that number found no sign that repairs fell behind. The source is a Project Glasswing preprint, reported by CryptoSlate on November 10, 2026.
What the study says
The paper links AI-assisted work to a sharp jump in disclosures. And the fixes? By the researchers' own measure, they kept pace with the surge. Nothing in the summary suggests repair times slowed down.
So far, so good. Now the catch.
What it doesn't settle
The summary flags two open problems. One is hidden backlogs. A repair rate can look steady in the measured data while a pile of unfixed bugs sits underneath it, uncounted. The other is attack prevention. More disclosures and quick patches don't prove fewer real-world exploits. Finding bugs isn't the same as stopping attackers.
It's also a preprint, which means no formal peer review yet. I didn't have the details of method and scope, so I'd hold that headline figure loosely.
Why it matters
Crypto runs on code that holds money directly. A smart contract bug, a wallet flaw or an exchange vulnerability turns into real losses fast. If AI tools help teams find more problems without drowning the people who fix them, that's good news for anyone with funds in DeFi protocols or on exchanges.
But a rising disclosure count isn't a safety score. I'd look for audits that report fix times and unresolved issues, not just bugs found. The real question is whether AI helps defenders faster than it helps attackers.
Source: CryptoSlate
Top Crypto Casinos Right Now
🇯🇵 Showing sites that accept players from Japan Change country
Advertising disclosure: we may earn a commission when you join an operator via links on this page, at no cost to you. This never dictates our ratings. How we make money
More info Less info
Why we chose it: A US-facing book that pays a premium for crypto deposits: 125% against 100% for cash, with fee-free deposits across five coins.
- Payout time
- Within 24 hours (crypto)
- KYC
- Required
Restricted countries: Australia, Afghanistan, Bulgaria, Central African Republic, Congo - Brazzaville, Eritrea, France, Guinea-Bissau and 16 more.
More info Less info
Why we chose it: Phemex is a derivatives-first venue for Europe, Asia and Latin America. Good engine, mid-size liquidity, and a blocklist that grew in 2025.
- Online since
- 2019
- Licence
- Seychelles
- KYC
- Required
Restricted countries: United Arab Emirates, Afghanistan, Australia, Belgium, Bermuda, Belarus, Ontario (Canada), China and 18 more.
More info Less info
Why we chose it: BetNow's casino floor with the biggest match bonus we list. The 200% headline deserves a careful read of the code terms behind it.
- Payout time
- Within 24 hours (crypto)
- KYC
- Required
Restricted countries: Australia, Afghanistan, Bulgaria, Central African Republic, Congo - Brazzaville, Eritrea, France, Guinea-Bissau and 16 more.
Priya Nair
Priya covers the AI side of crypto — agent tokens, decentralised compute and where the two industries actually meet.
