20 place 0

1016 AI Evaluators Struggle with Models That Know When They’re Being Tested

The Information
Rocket Drew @ The Information · 06/01/2026 10:00 EDT

AI researchers are starting to make progress on a confounding problem: AI models are getting better at telling when they are in an evaluation.That could become a problem for AI companies that use evaluations to gauge the capabilities and behaviors of their models before releasing them. If models act differently during testing, that could mean they get released with undesirable tendencies. It could also undermine their creators’ ability to show off test scores to potential clients. Evaluations are important.

To see detailed statistics for the news please log in »

Read the original

Add your comment
You must be logged in with Facebook to read and write comments.

A newsletter a day!

You may get 10 most important news around midday in daily newsletter. Press the button and we will send you the most important news only, no spam attached.

or register

LIKE us on Facebook so you won't miss the most important news of the day!

News from the same source
The Information The Information
Silicon Valley
George Avalos @ Silicon Valley 1 place · 02/07/2106 01:28 EDT

Newark apartment complex bought for much less than prior value

An East Bay apartment complex has been bought at a price that's well below its prior value. Read more â€ș

0

🔼
21.06.2026 ♋ Dear Cancer, today you can expect a day filled with a variety of emotions and... Read more â€ș
Silicon Valley
George Avalos @ Silicon Valley 2 place · 02/07/2106 01:28 EDT

PG&E buys San Jose building to bolster South Bay operations

A PG&E Corp. unit has bought a San Jose building in a move to bolster the utility's South Bay operations. Read more â€ș

0

Digital Trends
Shimul Sood @ Digital Trends 1 place · today 15:44 EDT

Hackers leak facial recognition records tied to millions of Madison Square Garden visitors

A cybercriminal group has published what it claims are millions of records stolen from Madison Square Garden Entertainment. The leak is drawing attention not just because of its size, but because it includes facial recognition data, internal threat assessments, and detailed visitor profiles. Read more â€ș

0 newcommer

Eurogamer.net
Vikki Blake @ Eurogamer.net 1 place · today 15:42 EDT

CD Projekt Red co-CEO admits it "indefinitely" "lost the faith" of some fans after Cyberpunk 2077

The Witcher 4 developer CD Projekt Red believes it's yet to complete its "full redemption arc" after the disastrous 2020 launch of its open-world action-adventure game, Cyberpunk 2077. Read more Read more â€ș

0 newcommer

Slashdot
EditorDavid @ Slashdot 1 place · today 15:40 EDT

Cops Keep Getting Arrested for Using Flock's Cameras to Stalk People

404 Media remembers how a Florida police office looked up his ex-girlfriend's license plate in the Flock automated license plate reader system at least 69 times in 2024 — even searching for her mom's license plate at least 24 times. The police office was charged with stalking and hacking-related offenses, serving one day in prison with five years of probation — but his case "was not a one-off." [Alternate link... Read more â€ș

0 newcommer

SlashGear
SlashGear 1 place · today 15:15 EDT

Why Do Utah State Highway Signs Have A Beehive?

Utah's highway signs stand out from those used in most states, and the unusual symbol is a reflection of an important part of the state's history. Read more â€ș

0 fresh

Habr
StudyQA @ Habr 1 place · today 15:12 EDT

АуЮот Đ°Đ»ĐłĐŸŃ€ĐžŃ‚ĐŒĐŸĐČ: ĐșаĐș Ń€Đ”Đ°Đ»ĐžĐ·Đ°Ń†ĐžŃ Boyer-Moore с 190K Đ·ĐČёзЎ ĐœĐ° GitHub ĐŸĐșĐ°Đ·Đ°Đ»Đ°ŃŃŒ brute-force

ĐŸŃ€ĐŸĐČДрОл рДалОзацОю Boyer-Moore ĐČ TheAlgorithms/Python (190K+ Đ·ĐČёзЎ). ОĐșĐ°Đ·Đ°Đ»ĐŸŃŃŒ, Ń‡Ń‚ĐŸ сЎĐČОг bad character запОсыĐČĐ°Đ”Ń‚ŃŃ ĐČ ĐżĐ”Ń€Đ”ĐŒĐ”ĐœĐœŃƒŃŽ for-цоĐșла, Ń‡Ń‚ĐŸ ĐČ Python ĐœĐ” ĐžĐŒĐ”Đ”Ń‚ ŃŃ„Ń„Đ”Đșта. ĐĐ»ĐłĐŸŃ€ĐžŃ‚ĐŒ ĐČыЮаёт праĐČĐžĐ»ŃŒĐœŃ‹Đ” Ń€Đ”Đ·ŃƒĐ»ŃŒŃ‚Đ°Ń‚Ń‹, ĐœĐŸ Ń€Đ°Đ±ĐŸŃ‚Đ°Đ”Ń‚ ĐșаĐș brute-force O(nm) ĐČĐŒĐ”ŃŃ‚ĐŸ O(n/m). ĐŸĐ»ŃŽŃ Дщё ĐŽĐČĐ” ĐœĐ°Ń…ĐŸĐŽĐșĐž: бДсĐșĐŸĐœĐ”Ń‡ĐœŃ‹Đč цоĐșĐ» ĐČ Ń‚ĐžĐżĐžŃ‡ĐœŃ‹Ń… Ń€Đ”Đ°Đ»ĐžĐ·Đ°Ń†ĐžŃŃ… full BM Đž ĐŸŃˆĐžĐ±Đșа ĐČ ĐŸŃ€ĐžĐłĐžĐœĐ°Đ»ŃŒĐœĐŸĐč ŃŃ‚Đ°Ń‚ŃŒĐ” 1977 ĐłĐŸĐŽĐ°, ĐșĐŸŃ‚ĐŸŃ€ŃƒŃŽ оспраĐČОлО Ń‚ĐŸĐ»ŃŒĐșĐŸ ĐČ 1980-ĐŒ. Чотать ЎалДД Read more â€ș

0 fresh

Digital Trends
Shimul Sood @ Digital Trends 2 place · today 15:08 EDT

Thanks to AI, a Chinese startup has figured out the priciest fusion energy bottleneck

Fusion energy has spent decades trapped in an expensive cycle of trial and error. Now, a Chinese startup believes AI-powered simulation software could dramatically accelerate reactor development by helping scientists test designs virtually before committing to costly real-world experiments. Read more â€ș

0 fresh

TechRadar
TechRadar 1 place · today 15:05 EDT

How to watch Uruguay vs Cape Verde: Free Streams & TV Channels for FIFA World Cup 2026

Here's how to watch Uruguay vs Cape Verde for free online and from anywhere as World Cup 2026 underdogs Cape Verde look to spring another surprise. Read more â€ș

0 fresh

CNET
Nasha Addarich Martínez @ CNET 1 place · today 15:00 EDT

The Best LED Face Masks That Will Improve Your Skin's Appearance

We tested popular FDA-cleared LED face masks to find the best ones for your home needs. Read more â€ș

0 newcommer

The Verge
Terrence O’Brien @ The Verge 1 place · today 14:53 EDT

Bose thinks it can be a media company for some reason

The history books are littered with the corpses of corporate record labels started by companies that had no business being in the music industry. Bose thinks it can be the exception to the rule. It thinks it can be Red Bull. And, while Bose has more of a right to dip its toes into the [
] Read more â€ș

0 fresh

Business Insider
Kelly Burch @ Business Insider 1 place · today 14:51 EDT

I left the Navy SEALs to have more time with my 3 kids. What I learned in the military helped me raise confident kids.

Former Navy SEAL Brandon Webb says lessons from sniper training helped him teach his children confidence, resilience, and independence. Read more â€ș

0 fresh

Gizmodo
Justin Carter @ Gizmodo 1 place · today 14:50 EDT

Marvel’s New Comics Universe Is Starting All at Once

Marvel wants its new 'Midnight' books to feel like a big deal, so they're getting a full week all to themselves. Read more â€ș

0 fresh

Habr
Akhmadaliev @ Habr 2 place · today 14:47 EDT

RICE, ICE, MoSCoW: ĐșĐŸĐłĐŽĐ° фрДĐčĐŒĐČĐŸŃ€Đș ĐżŃ€ĐžĐŸŃ€ĐžŃ‚ĐžĐ·Đ°Ń†ĐžĐž ĐČас Ń‚ĐŸĐżĐžŃ‚

ĐšĐŸĐłĐŽĐ° я ĐżŃ€ĐžŃˆŃ‘Đ» ĐČ Instameal, у ĐœĐ°Ń был Đ±ŃĐșĐ»ĐŸĐł ĐœĐ° ŃĐŸŃ€ĐŸĐș заЎач Đž ĐœĐž ĐŸĐŽĐœĐŸĐłĐŸ чётĐșĐŸĐłĐŸ ĐșŃ€ĐžŃ‚Đ”Ń€ĐžŃ ĐżĐŸŃ‡Đ”ĐŒŃƒ ĐŸĐŽĐœĐŸ ĐČĐ°Đ¶ĐœĐ”Đ” ĐŽŃ€ŃƒĐłĐŸĐłĐŸ.Мы ĐżĐŸĐżŃ€ĐŸĐ±ĐŸĐČалО RICE. ĐŸĐŸŃ‚ĐŸĐŒ ICE. ĐŸĐŸŃ‚ĐŸĐŒ MoSCoW. ĐŸĐŸŃ‚ĐŸĐŒ ŃĐœĐŸĐČа RICE с ĐŽŃ€ŃƒĐłĐžĐŒĐž ĐČĐ”ŃĐ°ĐŒĐž.ĐŸŃ€ĐŸĐ±Đ»Đ”ĐŒĐ° была ĐœĐ” ĐČ Ń‚ĐŸĐŒ, Ń‡Ń‚ĐŸ ĐŒŃ‹ ĐČыбОралО ĐœĐ”ĐżŃ€Đ°ĐČĐžĐ»ŃŒĐœŃ‹Đč фрДĐčĐŒĐČĐŸŃ€Đș. ĐŸŃ€ĐŸĐ±Đ»Đ”ĐŒĐ° была ĐČ Ń‚ĐŸĐŒ, Ń‡Ń‚ĐŸ ĐŒŃ‹ ĐŽŃƒĐŒĐ°Đ»Đž: ĐČŃ‹Đ±Đ”Ń€Đ”ĐŒ праĐČĐžĐ»ŃŒĐœŃ‹Đč ĐžĐœŃŃ‚Ń€ŃƒĐŒĐ”ĐœŃ‚ - Đž ĐżŃ€ĐžĐŸŃ€ĐžŃ‚Đ”Ń‚Ń‹ ĐČŃ‹ŃŃ‚Ń€ĐŸŃŃ‚ŃŃ ŃĐ°ĐŒĐž.ĐĐ” ĐČŃ‹ŃŃ‚Ń€ĐŸŃŃ‚ŃŃ.Đ§Ń‚ĐŸ таĐșĐŸĐ” ĐșажЎыĐč Оз трёхRICE: Reach (ĐŸŃ…ĐČат) × Impact (ĐČĐ»ĐžŃĐœĐžĐ”) × Confidence (уĐČĐ”Ń€Đ”ĐœĐœĐŸŃŃ‚ŃŒ)... Read more â€ș

0 fresh

SlashGear
SlashGear 2 place · today 14:45 EDT

Is The Leatherman Arc Worth The Price? Owners Have This To Say About It

The Leatherman Arc is one of the most expensive multitools on the market, so it's no surprise that owners and reviewers have strong opinions about its value. Read more â€ș

0 fresh

Habr
GlobalSign_admin (GlobalSign) @ Habr 3 place · today 14:37 EDT

ХДрĐČосы ĐșĐŸĐœĐČДртацОО ĐșĐŸĐŽĐ° «съДЎают» ĐŸĐżĐ”ĐœŃĐŸŃ€Ń

ĐĐ”ĐŽĐ°ĐČĐœĐŸ ĐČ ĐžĐœŃ‚Đ”Ń€ĐœĐ”Ń‚Đ” ĐœĐ°Ń‡Đ°Đ» Ń€Đ°Đ±ĐŸŃ‚Ńƒ сДрĐČОс рДфаĐșŃ‚ĐŸŃ€ĐžĐœĐłĐ° Malus.sh ĐżĐŸ Â«ĐŸŃ‡ĐžŃŃ‚ĐșĐ” ĐșĐŸĐŽĐ° ĐŸŃ‚ ĐŸĐżĐ”ĐœŃĐŸŃ€ŃĐœŃ‹Ń… Đ»ĐžŃ†Đ”ĐœĐ·ĐžĐč». ĐžĐœ ĐżĐŸĐ·ĐžŃ†ĐžĐŸĐœĐžŃ€ŃƒĐ”Ń‚ ŃĐ”Đ±Ń ĐșаĐș Â«Ń‡ĐžŃŃ‚Đ°Ń ĐșĐŸĐŒĐœĐ°Ń‚Đ°Â», гЎД ŃĐŸŃ„Ń‚ ĐŸŃ‡ĐžŃ‰Đ°Đ”Ń‚ŃŃ ĐŸŃ‚ Đ»ĐžŃ†Đ”ĐœĐ·ĐžĐŸĐœĐœĐŸĐłĐŸ Đ±Ń€Đ”ĐŒĐ”ĐœĐž. йуЎа Đ·Đ°ĐłŃ€ŃƒĐ¶Đ°Đ”Ń‚ŃŃ ĐŒĐ°ĐœĐžŃ„Đ”ŃŃ‚ сĐČĐŸĐ±ĐŸĐŽĐœĐŸĐłĐŸ ĐżŃ€ĐŸĐ”Đșта, а LLM за ĐœĐ”Đ±ĐŸĐ»ŃŒŃˆŃƒŃŽ ĐżĐ»Đ°Ń‚Ńƒ пДрДпОсыĐČаДт ĐșĐŸĐŽ с ŃĐŸŃ…Ń€Đ°ĐœĐ”ĐœĐžĐ”ĐŒ Ń„ŃƒĐœĐșŃ†ĐžĐŸĐœĐ°Đ»ŃŒĐœĐŸŃŃ‚Đž. Đ˜ĐŽĐ”Ń ĐČ Ń‚ĐŸĐŒ, Ń‡Ń‚ĐŸ ĐœĐŸĐČыĐč ĐșĐŸĐŽ ĐŒĐŸĐ¶ĐœĐŸ ĐžŃĐżĐŸĐ»ŃŒĐ·ĐŸĐČать ĐșаĐș ŃƒĐłĐŸĐŽĐœĐŸ, бДз ŃĐŸĐ±Đ»ŃŽĐŽĐ”ĐœĐžŃ Ń‚Ń€Đ”Đ±ĐŸĐČĐ°ĐœĐžĐč сĐČĐŸĐ±ĐŸĐŽĐœŃ‹Ń… Đ»ĐžŃ†Đ”ĐœĐ·ĐžĐč APGL, MIT, Apache Đž Юр., ĐżĐŸĐŽ ĐșĐŸŃ‚ĐŸŃ€Ń‹ĐŒĐž ĐŸĐżŃƒĐ±Đ»ĐžĐșĐŸĐČĐ°Đœ ĐŸŃ€ĐžĐłĐžĐœĐ°Đ».ĐĐ”ĐŽĐŸĐ±Ń€ĐŸŃĐŸĐČĐ”ŃŃ‚ĐœŃ‹Đ” Ń€Đ°Đ·Ń€Đ°Đ±ĐŸŃ‚Ń‡ĐžĐșĐž ĐżĐŸĐ»ŃƒŃ‡Đ°ŃŽ Read more â€ș

0 fresh

Habr
TatarnikovEgor @ Habr · today 14:32 EDT

ĐšĐŸĐłĐŽĐ° Đ»ŃƒŃ‡ŃˆĐ” публОĐșĐŸĐČаться ĐœĐ° ЄабрД. ХтатОстОчДсĐșĐžĐč Đ°ĐœĐ°Đ»ĐžĐ· сĐČŃĐ·Đž ĐČŃ€Đ”ĐŒĐ”ĐœĐž публОĐșацоо Đž ĐŸŃ…ĐČата статДĐč

На ЄабрД сДĐčчас ĐČŃ‹ŃĐŸĐșая ĐșĐŸĐœĐșŃƒŃ€Đ”ĐœŃ†ĐžŃ срДЎО аĐČŃ‚ĐŸŃ€ĐŸĐČ Đ·Đ° ĐČĐœĐžĐŒĐ°ĐœĐžĐ” чОтатДлДĐč. ĐŸĐŸ ĐŽĐ°ĐœĐœŃ‹ĐŒ ŃĐ°ĐŒĐŸĐłĐŸ Єабра, ĐČ 2025 ĐłĐŸĐŽŃƒ ĐœĐ° саĐčтД Đ±Ń‹Đ»ĐŸ Đ±ĐŸĐ»Đ”Đ” 10 тысяч ŃƒĐœĐžĐșĐ°Đ»ŃŒĐœŃ‹Ń… аĐČŃ‚ĐŸŃ€ĐŸĐČ ĐșĐŸĐœŃ‚Đ”ĐœŃ‚Đ°, а ĐșĐŸĐ»ĐžŃ‡Đ”ŃŃ‚ĐČĐŸ публОĐșацоĐč прДĐČŃ‹ŃĐžĐ»ĐŸ 51 тысячу. Đ­Ń‚ĐŸ ĐŸĐ·ĐœĐ°Ń‡Đ°Đ”Ń‚, Ń‡Ń‚ĐŸ ЎажД ĐșачДстĐČĐ”ĐœĐœŃ‹Đč ĐŒĐ°Ń‚Đ”Ń€ĐžĐ°Đ» ĐŒĐŸĐ¶Đ”Ń‚ ĐœĐ” ĐżĐŸĐ»ŃƒŃ‡ĐžŃ‚ŃŒ Đ·Đ°ĐŒĐ”Ń‚ĐœŃ‹Đč ĐŸŃ…ĐČат Оз-за Đ±ĐŸĐ»ŃŒŃˆĐŸĐłĐŸ ĐșĐŸĐ»ĐžŃ‡Đ”ŃŃ‚ĐČа публОĐșацоĐč ĐČ Đ»Đ”ĐœŃ‚Đ”.Есть Ń€Đ°ŃĐżŃ€ĐŸŃŃ‚Ń€Đ°ĐœŃ‘ĐœĐœĐŸĐ” ĐŒĐœĐ”ĐœĐžĐ”, Ń‡Ń‚ĐŸ публОĐșĐŸĐČать статьо ĐœŃƒĐ¶ĐœĐŸ ĐČ ĐżŃ€Đ”ĐŽĐŸĐ±Đ”ĐŽĐ”ĐœĐœĐŸĐ” ĐČŃ€Đ”ĐŒŃ, Ń‡Ń‚ĐŸĐ±Ń‹ люЎО ĐœĐ° ĐŸĐ±Đ”ĐŽĐ”ĐœĐœĐŸĐŒ пДрДрыĐČĐ” ĐŒĐŸĐłĐ»Đž ĐżĐŸŃ‡ĐžŃ‚Đ°Ń‚ŃŒ это статьо, Ń‚ĐŸĐłĐŽĐ° ĐŸŃ…ĐČат Đ±ŃƒĐŽĐ”Ń‚... Read more â€ș

0 fresh

The most popular news from the same source for the last week
The Information The Information
The Information
Abram Brown @ The Information · 06/14/2026 16:07 EDT

Demand for knockoff versions of retatrutide, a peptide-based drug being developed by Eli Lilly, has skyrocketed lately, propelling a gray market that likely exceeds $100 million in annual sales, according to an estimate prepared by The Information that examined many of the market’s biggest ... Read more â€ș

0

The Information
Martin Peers @ The Information · 06/14/2026 18:01 EDT

What a weekend! Set time aside this coming Thursday for the New York City ticker tape parade celebrating the victorious Knicks. The joy flowing from that game Saturday night tops even the  plaudits flowing from SpaceX’s successful IPO (read down for more on that subject.) Still, SpaceX was quickly supplanted in the business limelight by the latest Anthropic-government drama, this time over the AI firm’s recently released Fable 5 model.... Read more â€ș

0

The Information
Jing Yang @ The Information · 06/15/2026 05:35 EDT

Tencent Holdings has backed a new AI lab founded by Junyang Lin, former lead researcher of Alibaba Group’s Qwen models, according to two people with knowledge of the investment. Tencent invested $20 million in Lin’s first funding round that raised several hundred million dollars at a post-money ... Read more â€ș

0

The Information
Phoebe Liu @ The Information · 06/15/2026 06:55 EDT

Data center software startup and AI-server broker Hydra Host has raised $100 million at a valuation of close to $800 million, led by Kindred Ventures. Nvidia, Cathie Wood’s ARK Invest, early CoreWeave backer Magnetar, and existing investors Founders Fund and Flume Ventures also participated. ... Read more â€ș

0

The Information
Martin Peers @ The Information · 06/15/2026 07:51 EDT

The Murdoch family’s Fox is buying Roku for $22 billion in cash and stock, or $160 a share, giving Roku shareholders an exit at the highest price at which the stock has traded since the 2020-2022 Covid-fueled surge. In recent years Roku has traded mostly below $100 a share. Roku, which sells ... Read more â€ș

0

The Information
Dakin Campbell @ The Information · 06/15/2026 09:00 EDT

When Broadcom last week announced a funding venture with Apollo and Blackstone to pay for a gigawatt of computing capacity to be used by Anthropic, it looked like the latest in a series of AI computing deals funded by private equity. Behind the scenes, however, the deal represents a risky move by Broadcom to boost demand for its chips. While the announcement didn’t spell out Broadcom’s role, the company—which works... Read more â€ș

0

The Information
Valida Pau @ The Information · 06/15/2026 09:35 EDT

Salesforce has agreed to buy Fin, a startup that develops customer agents formerly known as Intercom, for $3.6 billion, as the software giant hopes to win new businesses from enterprises to adopt its own AI offering. The sale price is a big premium to Fin’s last valuation of $2 billion set at a ... Read more â€ș

0

The Information
Qianer Liu @ The Information · 06/15/2026 10:01 EDT

Anthropic can’t catch a break! Leaders of the AI firm are meeting with Trump administration officials in D.C. this week to try to resolve a dispute that blew up on Friday night. The administration said that it would need to cut off access to its most advanced Claude Fable 5 and Claude Mythos 5 models to foreign nationals, which led to the company taking down the models altogether. Then, on... Read more â€ș

0

The Information
Phoebe Liu @ The Information · 06/15/2026 11:31 EDT

As AI developers and cloud providers have launched server chips to lessen their dependence on Nvidia’s, some analysts and executives at these firms expected the chips to eat into Nvidia’s market share.That doesn’t seem to be happening.Nvidia has actually increased its share of the market for chips that power existing AI models, known as inference, to 74% from 66% over the past year, according to The Information’s estimates based on... Read more â€ș

0

The Information
Stephanie Palazzolo @ The Information · 06/15/2026 12:04 EDT

A Washington, D.C.-based Anthropic customer filed a class action lawsuit against Anthropic Sunday night alleging that the company had misled customers about the value of its premium “Max 5x” and “Max 20x” subscription plans. The lawsuit alleges that Anthropic markets its Max 20x plan, which ... Read more â€ș

0

Most popular sources

  • You see 305 news out of 305.
  • Sources 61 out of 61.
Silicon Canals 0%
Inc42 Media 0%
Wired 0%
The Fintech Times 0%
Vox 0%
View sources »

LIKE us on Facebook so you won't miss the most important news of the day!

21.06.2026 15:51
Last update: 15:45 EDT.
News rating updated: 22:40.

What is Times42?

Times42 brings you the most popular news from tech news portals in real-time chart.
Read about us in FAQ section.


Times42 © 2026