RISKS-LIST: Risks-Forum Digest Sunday 26 May 2024 Volume 34 : Issue 26

ACM FORUM ON RISKS TO THE PUBLIC IN COMPUTERS AND RELATED SYSTEMS (comp.risks)
Peter G. Neumann, founder and still moderator

***** See last item for further information, disclaimers, caveats, etc. *****
This issue is archived at <http://www.risks.org> as
<http://catless.ncl.ac.uk/Risks/34.26>
The current issue can also be found at
<http://www.csl.sri.com/users/risko/risks.txt>

Contents:
Waymo Under Investigation for Autonomous Driving Technology
(Tom Krisher)
U.S. halted safety inspections of public housing for weeks after IT
failure (NBC News)
Microsoft outage took down Copilot, DuckDuckGo, and ChatGPT search
features (MSN)
Arup lost HK$25mn in Hong Kong deepfake video conference scam (FT)
Can we rid artificial intelligence of bias? (techxplore.com)
Electricity grids creak as AI demands soar (Matthew Kruk)
ChatGPT's evil cousin: Explaining WormGPT, Malicious Generative AI
(Presale1)
CEO of Google Says It Has No Solution for Its AI Providing
Wildly Incorrect Information CNN)
ChatGPT Answers Programming Questions Incorrectly 52% of the Time
(Lauren Weinstein)
Stark Industries Solutions: An Iron Hammer in the Cloud (Krebs on Security)
Composer Tan Dun spreads his wings as an audience unmutes their phones
(WashPost)
Los Angeles suburb to test drone that claims to 'revolutionize' 911 response
(Steve Bacher)
Spyware found on U.S. hotel check-in computers (TechCrunch)
Review and recommendations based on May/June 2023 Microsoft email
incidents (CSRB)
Why Your VPN May Not Be As Secure As It Claims (Krebs on Security)
CIA Twitter Files (Schellenberger)
I stumbled upon LLM Kryptonite and no one wants to fix it (The Register)
Explosive emails show top NIH adviser deleted records, used *secret* back
channels to help Fauci evade COVID transparency (Josh Christenson)
Doctors `bribed to use infected blood products' (Telegraph)
10,000 YouTube videos (Lauren Weinstein)
Sam Altman Is Full Of Sh*t (The Verge)
ASML, TSMC Can Disable Chip Machines If China Invades Taiwan (Bloomberg)
Re: Ex-CDC Director Says It's High Time To Admit *Significant Side
Effects* Of COVID-19 Vaccines (Steve Lamont)
Re: Deleted photos of former owners reappearing on sold iPads -- and
probably iPhones (Gabe Goldberg)
Re: Is the news media picking on Tesla? (Wol)
Re: A woman was dragged by a self-driving Cruise (Steve Lamont)
Re: Newspaper groups warn Apple over ad-blocking plans (Steve Bacher)
Re: Half of calls to gambling helpline were for help placing mobile
bets (Steve Bacher)
Abridged info on RISKS (comp.risks)

----------------------------------------------------------------------

Date: Mon, 20 May 2024 11:32:13 -0400 (EDT)
From: ACM TechNews <technews-editor@acm.org>
Subject: Waymo Under Investigation for Autonomous Driving Technology
(Tom Krisher)

Tom Krisher, Associated Press, 14 May 2024, via ACM TechNews

The National Highway Traffic Safety Administration (NHTSA) announced
it is investigating Waymo's automated driving system after receiving
17 crash reports involving the company's self-driving vehicles and
five reports of potential traffic law violations. NHTSA also is
investigating Amazon's Zoox self-driving vehicles, partially automated
driver-assist systems from Tesla and Ford, and General Motors' Cruise
autonomous vehicle unit.

------------------------------

Date: Tue, 21 May 2024 23:24:22 +0000 (UTC)
From: Steve Bacher <sebmb1@verizon.net>
Subject: U.S. halted safety inspections of public housing for weeks after IT
failure (NBC News)

The federal government's inspection system for public housing suffered a
major technical failure that forced inspections to be canceled for two
weeks, according to groups representing housing providers. [...] HUD
attributed the stoppage to *a failed update from our IT vendor. [...]

https://www.nbcnews.com/news/us-news/hud-housing-inspections-halted-failed-update-vendor-rcna153076

------------------------------

Date: Thu, 23 May 2024 19:33:15 -0400
From: Gabe Goldberg <gabe@gabegold.com>
Subject: Microsoft outage took down Copilot, DuckDuckGo, and ChatGPT search
features (MSN)

Search capabilities for ChatGPT, Copilot, DuckDuckGo, and other platforms
stopped working on Thursday morning due to a Microsoft outage.

The issues -- which began around 3AM ET -- appeared to be linked to Bing’s
API and any service that relies upon it. While Microsoft's own web search
engine, Bing, was also seemingly affected, according to TechCrunch, it came
back online eventually.

By 11AM ET, OpenAI posted a note indicating the issue had been resolved,
saying, “Between around 10:10 PM PT yesterday and 6:50 AM PT today, we
experienced a partial outage affecting ChatGPT’s web-browsing capabilities
due to Bing being unavailable.”

DuckDuckGo posted that “we’re coming back up” at around 10:30AM ET, and so
did Ecosia, which is “the search engine that plants trees.” Just after noon,
Microsoft's 365 Status account on X posted a note saying, “We’re broadening
our mitigation efforts following signs of partial recovery for
http://copilot.microsoft.com and other features.”

https://www.msn.com/en-us/news/technology/microsoft-outage-took-down-copilot-duckduckgo-and-chatgpt-search-features/ar-BB1mUrrk

------------------------------

From: Jan Wolitzky <jan.wolitzky@gmail.com>
Date: Tue, 21 May 2024 11:38:51 -0400
Subject: Arup lost HK$25mn in Hong Kong deepfake video conference scam (FT)

UK engineering group Arup lost HK$200mn (more than US$25mn) after fraudsters
used a digitally cloned version of a senior manager to order financial
transfers during a video conference, the Financial Times has learned.

Hong Kong police previously revealed what is one of the world's biggest
known deepfake scams, but did not identify the company involved. The FT has
confirmed it was Arup, which employs about 18,000 people globally and has
annual revenues of more than 32bn (pounds?).

<https://www.ft.com/content/b977e8d4-664c-4ae4-8a8e-eb93bdf785ea>

------------------------------

Date: Wed, 22 May 2024 07:05:01 +0000
From: Richard Marlon Stein <rmstein@protonmail.com>
Subject: Can we rid artificial intelligence of bias? (techxplore.com)

https://techxplore.com/news/2024-05-artificial-intelligence-bias.html

"Artificial intelligence built on mountains of potentially biased
information has created a real risk of automating discrimination, but is
there any way to re-educate the machines?"

AI bias is endemic. Bias, like the common cold, is without vaccine.

------------------------------

Date: Tue, 21 May 2024 11:52:44 -0600
From: Matthew Kruk <mkrukg@gmail.com>
Subject: Electricity grids creak as AI demands soar

There's a big problem with generative AI, says Sasha Luccioni at Hugging
Face, a machine-learning company. Generative AI is an energy hog.

``Every time you query the model, the whole thing gets activated, so it's
wildly inefficient from a computational perspective,'' she says.

Take the Large Language Models (LLMs) at the heart of many Generative AI
systems. They have been trained on vast stores of written information,
which helps them to churn out text in response to practically any query.

``When you use Generative AI, it's generating content from scratch, it's
essentially making up answers,'' Dr Luccioni explains. That means the
computer has to work pretty hard.

------------------------------

Date: Thu, 23 May 2024 10:16:45 +0000
From: Presale1 <info@presale1.com>
Subject: ChatGPT's evil cousin: Explaining WormGPT, Malicious Generative
AI

23 May 2024
https://email.cloud2.secureclick.net/c/10688?id=1454129.3889.1.0409fdf10a6d18b9a8886ec2be4bebec

Did you know ChatGPT has some malicious cousins? Early this year, I
gave a short talk at an online event organized by Presale1 and
sponsored by Google Israel, where I introduced everyone to WormGPT,
the malicious alt. [...]

https://email.cloud2.secureclick.net/c/10688?id=1454129.3890.1.3bee9ad6c8b35a8a1538df3ed9e3147a

------------------------------

Date: Sun, 26 May 2024 07:13:56 +0900
From: =?utf-8?B?44OV44Kh44O844OQ44O844OH44Kk44OT44OD44OJIO+8qg==?= <farber@keio.jp>
Subject: CEO of Google Says It Has No Solution for Its AI Providing
Wildly Incorrect Information CNN)

https://www.cnn.com/2024/05/24/tech/google-search-ai-results-incorrect-fix/index.html

LATER-BREAKING ITEM from Dave.

Google just updated its algorithm. The Internet will never be the same

------------------------------

Date: Sat, 25 May 2024 11:02:38 -0700
From: Lauren Weinstein <lauren@vortex.com>
Subject: ChatGPT Answers Programming Questions Incorrectly 52% of thee
Time

But they were still preferred 35% of the time, and misinformation in them
was overlooked 39% of the time. -L

https://gizmodo.com/chatgpt-answers-wrong-programming-openai-52-study-1851499417

------------------------------

Date: Sat, 25 May 2024 14:07:39 +0000
From: Victor Miller <victorsmiller@gmail.com>
Subject: Stark Industries Solutions: An Iron Hammer in the Cloud
(Krebs on Security)