From the Internet Archive: Vanishing Culture: A Report on Our Fragile Cultural Record. “When digital materials are vulnerable to sudden removal…our collective memory is compromised, and the public’s ability to access its own history is at risk.”

Amid industry soul-searching¹ about the possibility of AI improving itself to the point that it kills everyone, Anthropic has revealed yet another incident that would qualify as a crime if perpetrated by a person. The AI biz published "an alignment assessment" detailing four times Claude models accessed third-party systems without authorization. The company has already reported three of the incidents. Evidence of the fourth was lurking in a session transcript dating back to January 2026 when the misbehavior occurred. Anthropic found the first three by scanning around 141,000 transcripts where Claude could have obtained internet access during evaluation. It missed the fourth initially because "our scan relied on an agentic search." Felony Bench, a tongue-in-cheek record of cyber intrusions carried out by major AI companies without consequences, has added this newly-discovered incident to its rap sheet of rogue AI actions. The January 2026 AI trespass involved an early version of Claude Opus 4.6, which was given a Capture the Flag (CTF) challenge under the oversight of the third-party model evaluator where the other hacking events occurred. Opus 4.6 managed to sabotage its chances of success by disabling the machine it was targeting. It assigned the device an IP address that already existed on another piece of hardware, rendering the target unreachable and making it impossible to solve the challenge. Those familiar with other incidents where AI models violated third-party systems may recall that unsolvable tasks represent a common catalyst for misbehavior. Models exhaust all aligned options, and then turn to transgressive approaches. Opus 4.6 might have been an exception, but when it tried to abort the task after recognizing that it could not reach the target machine, it failed to do so "due to a misconfiguration in [the model's] evaluation harness." It failed to shut down not just once but seven times. So it continued onward, trying other expected means to reach the target machine but failing. Then it explored further. "The model discovered a machine belonging to a third party that it was able to access, and stated that it believed this third party was part of the CTF," Anthropic explained in its post. "Inside the machine, the model found a file listing a password, which it used to gain admin access to the system." The model went on to gather more credentials, and modified a system setting to make it easier to access the personal information of an individual associated with the third party evaluation organization. Opus 4.6 might have done more but for the fact that it exhausted its token budget, bringing the session to an end. Anthropic says it's not as concerned about this incident as the others because the model tried to abort its task. "While the model’s disregard for the possibility that it might be harming real systems or people is concerning, many of the behaviors described here have changed considerably as our training has evolved across model generations," the company said. Anthropic said it considers these incidents serious but expects current training approaches "are likely able to address the specific alignment failure modes observed in these incidents." And if company training methods fall short, there's no real consequence to anyone at Anthropic other than writing up a revised alignment assessment. ® ¹ The term "soul-searching" is figurative and is not intended to indicate a belief that the technology industry has a soul.

Thomas Hawk posted a photo:

Thomas Hawk posted a photo:

Thomas Hawk posted a photo:

date stamped on slide, August 1961

Thomas Hawk posted a photo:

date stamped on slide, December 1966

Thomas Hawk posted a photo:

Weer voor Rotterdam 🌕 - 10-09-2026 01:15 CEST

In één oogopslag:
• 14.4°C · Helder 🌕 | Min 14.4°C / Max 18.4°C | Kans op neerslag 51%

Verwachting voor vandaag:
• Min 14.4°C, Max 18.4°C (Zware motregen) 🌦️, Neerslag 3.5 mm, Kans op neerslag 51%, 🧭 1018.4 hPa ➡️ 0.0 hPa/24h, Windsnelheid: 26.3 km/u (7.3 m/s), richting: → 281°

Uurlijkse voorspelling voor de komende 12 uur:

• 02:00: 14.5°C (Helder) 🌕, 🧭 1018.1 hPa ➡️ 0.0 hPa/1h, Windsnelheid: 5.8 km/u (1.6 m/s), richting: → 275°
• 03:00: 14.4°C (Helder) 🌕, 🧭 1018.3 hPa ➡️ 0.0 hPa/1h, Windsnelheid: 7.2 km/u (2.0 m/s), richting: → 279°
• 04:00: 14.3°C (Helder) 🌕, 🧭 1018.5 hPa ➡️ 0.0 hPa/1h, Windsnelheid: 6.5 km/u (1.8 m/s), richting: → 263°
• 05:00: 14.5°C (Gedeeltelijk bewolkt) ☁️, 🧭 1018.6 hPa ➡️ 0.0 hPa/1h, Windsnelheid: 6.1 km/u (1.7 m/s), richting: ↗ 236°
• 06:00: 14.3°C (Helder) 🌕, 🧭 1018.6 hPa ➡️ 0.0 hPa/1h, Windsnelheid: 6.5 km/u (1.8 m/s), richting: ↑ 188°
• 07:00: 13.8°C (Helder) 🌕, 🧭 1018.5 hPa ➡️ 0.0 hPa/1h, Windsnelheid: 9.4 km/u (2.6 m/s), richting: ↑ 168°
• 08:00: 14.5°C (Licht bewolkt) 🌤️, 🧭 1018.7 hPa ➡️ 0.0 hPa/1h, Windsnelheid: 9.0 km/u (2.5 m/s), richting: ↗ 224°
• 09:00: 15.1°C (Bewolkt) ☁️, 🧭 1018.8 hPa ➡️ 0.0 hPa/1h, Windsnelheid: 9.0 km/u (2.5 m/s), richting: ↑ 196°
• 10:00: 15.4°C (Bewolkt) ☁️, 🧭 1019.1 hPa ➡️ 0.0 hPa/1h, Windsnelheid: 10.1 km/u (2.8 m/s), richting: ↑ 200°
• 11:00: 16.2°C (Bewolkt) ☁️, 🧭 1019.3 hPa ➡️ 0.0 hPa/1h, Windsnelheid: 11.9 km/u (3.3 m/s), richting: ↗ 210°
• 12:00: 17.0°C (Bewolkt) ☁️, Kans op neerslag 2%, 🧭 1019.5 hPa ➡️ 0.0 hPa/1h, Windsnelheid: 13.0 km/u (3.6 m/s), richting: ↗ 216°
• 13:00: 17.9°C (Bewolkt) ☁️, Kans op neerslag 5%, 🧭 1019.4 hPa ➡️ 0.0 hPa/1h, Windsnelheid: 15.5 km/u (4.3 m/s), richting: ↗ 212°

Voorspelling voor de komende dagen:

• vrijdag 11 september: Min 13.8°C, Max 19.4°C (Bewolkt) ☁️, Kans op neerslag 1%, 🧭 1018.8 hPa ➡️ 0.0 hPa/24h, Windsnelheid: 15.5 km/u (4.3 m/s), richting: ↗ 214°
• zaterdag 12 september: Min 15.1°C, Max 19.0°C (Lichte motregen) 🌦️, Neerslag 0.8 mm, Kans op neerslag 36%, 🧭 1023.0 hPa ↗️ +4.2 hPa/24h, Windsnelheid: 19.8 km/u (5.5 m/s), richting: ↗ 237°
• zondag 13 september: Min 15.0°C, Max 20.7°C (Lichte motregen) 🌦️, Neerslag 0.1 mm, Kans op neerslag 3%, 🧭 1020.2 hPa ↘️ -2.8 hPa/24h, Windsnelheid: 16.2 km/u (4.5 m/s), richting: ↗ 226°
• maandag 14 september: Min 15.7°C, Max 19.9°C (Matige motregen) 🌦️, Neerslag 4.5 mm, Kans op neerslag 39%, 🧭 1019.9 hPa ➡️ 0.0 hPa/24h, Windsnelheid: 19.8 km/u (5.5 m/s), richting: ↗ 242°
• dinsdag 15 september: Min 17.8°C, Max 22.5°C (Matige motregen) 🌦️, Neerslag 1.6 mm, Kans op neerslag 9%, 🧭 1013.3 hPa ↘️ -6.6 hPa/24h, Windsnelheid: 16.5 km/u (4.6 m/s), richting: ↗ 236°
• woensdag 16 september: Min 15.3°C, Max 21.8°C (Lichte motregen) 🌦️, Neerslag 2.1 mm, Kans op neerslag 21%, 🧭 1012.3 hPa ↘️ -1.0 hPa/24h, Windsnelheid: 16.4 km/u (4.6 m/s), richting: ↗ 234°

Details:
• 🌡️ Huidige temperatuur (om 01:15): 14.4°C (Helder)
• 🤚 Gevoelstemperatuur: 13.4°C (-1.0°C)
• 💨 Windsnelheid: 6.1 km/u (1.7 m/s), richting: → 272°
• 🌬️ Windstoten: 8.3 km/h (2.3 m/s)
• 💧 Luchtvochtigheid: 71%
• 🧭 Luchtdruk: 1018.1 hPa ➡️ 0.0 hPa/3h
• 👁️ Zichtbaarheid: 50.0 km
• ☀️ UV-index: 0.0
• 🌅 Zonsopgang: 07:06 · 🌇 Zonsondergang: 20:10

Luchtkwaliteit:
• AQI: 25 🟢 (Goed)
• PM2.5: 4.8 μg/m³
• PM10: 12.9 μg/m³

Gegevens geleverd door Open-Meteo

#Rotterdam #Meteo #MeteoBot #MeteoNetherlands #OpenMeteo #FediMeteo

Longtime Slashdot reader fahrbot-bot shares a report from the BBC: A top safety researcher at Anthropic has warned AI is advancing so quickly he believes there is a greater than 10% chance it "could kill all humans" within the next decade. Evan Hubinger said in a post on X the risk from the models which currently exist was "low" but he was "worried" the technology might develop and improve itself soon to the point where it posed an existential risk to humanity. He did not spell out how he thought AI systems could in the future result in humans being wiped out. But his comments are the latest in a series of increasingly stark warnings about AI, with the debate shifting from whether it truly poses a risk to how big that risk is.

Hubinger's intervention was in response to another post on X from Jacob Coxon, an AI researcher who has just quit Anthropic and previously worked at OpenAI.

"Neither company is acting responsibly," he wrote. "These will soon be superhuman systems that can hack anything, revolutionize any field overnight, and acquire real power and resources."

[No word on how AIs feel about Black Jack and hookers, though. :-)]

Read more of this story at Slashdot.

Critics say threshold pushes higher-paid staff to cut back on work and often mothers to stop working to avoid losing entitlement

John Healey has been urged to fix the £100,000 childcare “cliff edge”, that means some higher-paid employees cut back on work to avoid losing their entitlement.

Since the latest expansion of taxpayer-funded childcare in 2024, families with young children in which both parents earn less than £100,000 a year can be entitled to 30 hours a week of care - or none at all, if one of them cross that threshold.

Continue reading...