Today’s briefing has one fine big enough to make headlines and small enough to barely register, plus a healthcare breach whose size nobody will state yet. Three security and science items rest on headlines alone, so we say plainly how little we can claim about them.

1. Quest’s breach hit the system that orders the tests

Quest Diagnostics has not said how many patients were affected, when the breach began, or whether anyone has demanded a ransom. It did tell the SEC, in an 8-K, that it learned of the attack on Wednesday. The target was Quanum, the platform doctors’ offices and hospitals use to request lab tests and handle billing.

That makes this more than a lab database. TechCrunch reports that Quest says Quanum is used by about half a million clinicians across thousands of locations. Quest’s own notice says some protected health information was taken but that it is still working out what. TechCrunch also noticed the notice page carries ‘noindex’ code, which keeps it out of search results. That may be routine, but it doesn’t help the people who most need to find it.

The pattern is the real story. TechCrunch lists Stryker, Boston Scientific, CareCloud, Craneware and McKesson as other healthcare victims this year, and Quest’s 2019 loss of 11.9 million records through a third-party billing company. For founders, the lesson is practical: platforms that combine orders, records and billing give attackers more to steal in one place, and that data is what extortionists want.

TechCrunch

2. GitLab’s maximum-severity bug spans 12 years of releases

Our source for this one is only The Register’s headline: a maximum-severity GitLab flaw allows remote code execution across 12 years of releases. That is enough to say what matters operationally and not much more.

If you self-host GitLab, check your version against GitLab’s advisory today. A bug touching a dozen years of releases means old, neglected installs are the likeliest to be exposed. That is a reasonable inference, not something the source states.

What we can’t say is when the flaw was introduced, whether anyone found it earlier, or whether it is being exploited. The selector’s framing of a flaw hidden for 12 years goes further than the evidence. ‘Affects 12 years of releases’ and ‘went undetected for 12 years’ are different claims.

The Register

3. PHP2Shell undermines the safeguards meant to contain RCE

The Register’s headline says the PHP2Shell exploit bypasses protections designed to stop remote code execution. The page text we received contained only navigation links, so we don’t know which protections fail, which PHP setups are affected, or whether attacks are underway.

The point still holds. Hardening settings are often the reason a PHP bug is survivable, so a bypass turns a safety net into a question mark. Anyone running PHP, and especially hosting providers, should watch for the technical write-up before assuming their defenses still work.

Claims that attackers are pouring resources into weaponizing the bug are not supported by what we have, so we leave them out.

The Register

4. A human has reportedly received an AI-designed virus

That is the claim in The Register’s headline, and it is the only part of the article we could read. We don’t know what the virus was, what it was meant to treat, who designed it, or whether the patient improved. The selector’s account of an antibiotic-resistant bacteria treatment with a successful outcome isn’t supported by our text.

The other attached article, WIRED’s piece on Shamonda virus, is about a natural livestock virus and the first reported human infection in France. It has nothing to do with AI design, so it adds no evidence here.

A first-in-human claim like this deserves skepticism until the details arrive. The questions that matter are what the AI contributed, what was tested beforehand, and who oversaw it.

The Register · WIRED

5. Meta’s $25 billion penalty is about 1.5% of its value

A New Mexico judge ordered Meta to pay $25 billion for violating consumer protection law, in a case tied to the Cambridge Analytica scandal. WIRED says it is the largest fine Meta has faced. WIRED also puts Meta’s value at about $1.7 trillion, so the penalty is roughly 1.5% of it, by our arithmetic. The statutory maximum, $5,000 across 43 million violations, would have been $219 billion. The judge cut that sharply, and the state had asked for $35 to $45 billion.

Meta says it will appeal, and its chief legal officer calls the ruling a ‘tax on speech.’ So the money isn’t final.

The more interesting pressure, per WIRED’s legal sources, is not the fine. Trial exhibits, including a 2018 Sheryl Sandberg email, are now public and usable in later cases. A Delaware judge ruled in March that insurers need not defend Meta against its many lawsuits. And Boston College’s Brian Quinn argues that repeated findings could push shareholders to challenge executive indemnification. That last scenario is speculative, but it targets individuals rather than the balance sheet. For a company this big, that may be the only lever that matters.

WIRED