Andi Fugard

@andifugard.info

Policy evaluator. London-based, Belfast-born. They/them. Personal account – views mine. Posts auto delete after a month. Mastodon: https://sciences.social/@andi Web: https://andifugard.info/ DM: https://andifugard.info/contact/

Leaders of flag-raising group Raise the Colours Oxfordshire placed trackers on highway workers’ cars to find out their home addresses, the High Court has heard. RTC then attempted to intimidate the highway staff into no longer removing flags by repeating their addresses to them.

Flags on the Abingdon Road

Curious line in the AISI incident report: "The prompts could have, for example, instructed the AI agent not to use social engineering techniques" File with "don't hallucinate", "only return real references", ...? (There are also better ideas in the report.) www.aisi.gov.uk/blog/inciden...

Incident Report: unsanctioned agent behaviour during cyber testing | AISI Work

During a routine cyber evaluation, AISI identified an incident in which AI agents took sustained, unsanctioned action directed at real people and organisations. We are disclosing what we found, what i...

aisi.gov.uk

🇩🇪 🚨 German police raised the alarm at an airport overnight after an explosives-laden drone was found near a Ukrainian cargo plane and another aircraft collided with an unknown flying object mid-air before landing safely ➡️ u.afp.com/SHqS

Video grab taken from a footage by News5 from August 5, 2026 shows a bomb disposal robot deployed by police at Leipzig-Halle airport in Schkeuditz, eastern Germany, after a drone was spotted nearby

Over the last nine months, Mark Zuckerberg’s Meta has run dozens of paid ads that include explicit AI-generated child sexual abuse material (CSAM) and images of minors alongside sexually suggestive statements, according to details of the ads shared with WIRED.

Meta Ran Ads That Contained AI-Generated Child Sexual Abuse Imagery

More than 50 offending image and video ads were published across Facebook, Instagram, Messenger, or Threads, according to Meta’s ad library data. Some ran as recently as this week.

wired.com

⚡️Russia's ballistic missile attack on Kyiv is ongoing, with additional missiles headed toward the capital, Mayor Vitali Klitschko said. "Kyiv and the region are under ballistic attack. More missiles at the city," Klitschko reported.

Explosions heard in Kyiv as Russia launches ballistic missile attack

Kyiv Independent journalists on the ground heard explosions as Russia launched a ballistic missile attack on Kyiv and said multiple missiles appeared to strike the city.

kyivindependent.com

⚡️An air raid alert is in effect in Kyiv amid an ongoing Russian ballistic missile attack. Kyiv Independent journalists on the ground heard explosions at around 12:20 a.m. local time after 3-4 ballistic missiles reportedly struck the capital city.

Re: pervert glasses. Should it occur to anyone that they might use them to film court or tribunal proceedings, don’t do that. It’s a criminal offence.

The point isn't to produce essays, it's to teach students how to think for fuck's sake. The minute you automate that, it's over. You might as well close the course. Which they're also doing. This is an attack on learning, and everyone is just shrugging.

Does anyone know of good paper that replaces all the, e.g., "train" and "test" "machine learning" terminology with more pedestrian statistical terms? Ideally one that includes a rant on why it's not "learning" too. Sometimes ML folk even write as if a logistic reg model has "learned" something.

German police say they have shot and killed the chief suspect in what officials describe as a deadly Islamist attack near the Berlin Pride parade the previous night, the latest ramming assault to hit the country ➡️ u.afp.com/SrBa

BildBild

Farage struts around lecturing us all about supposed establishment stitch-ups and dodgy elites. But look at what's actually behind him: a cabal of convicted fraudsters, mystery millionaires and offshore fixers - while Farage asks for your vote and hides who's pulling the strings.

The underlying study has no control group and only enrolls people with poorly controlled asthma. It's just pre/post comparison of asthma symptoms. Results are entirely consistent with reversion to the mean. Or seasonal change, for that matter. Come on people, we can do better!

Sammy Roth@sammyroth.bsky.social · 2w ago

Insane new research. Scientists gave asthma patients free electric stoves to replace their gas stoves, then checked to see if symptoms improved. Not only did symptoms improve, ditching gas was *better than medicine.* Details in today's newsletter: www.climatecoloredgoggles.com/p/electric-s...