RISKS-LIST: Risks-Forum Digest Tuesday 19 July 2022 Volume 33 : Issue 33

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/33.33>
The current issue can also be found at
<http://www.csl.sri.com/users/risko/risks.txt>

Contents:
The Big Hack: How China Used a Tiny Chip to Infiltrate U.S. Companies
(Bloomberg)
Driver says GPS made him turn onto train tracks in Everett; at least he was
able to escape before train destroyed his car (UniversalHub)
DeepMind AI Learns Simple Physics Like a Baby (Davide Castelvecchi)
As AI Language Skills Grow, So Do Scientists' Concerns (Matt O'Brien)
Researchers Defeat Facial Recognition Systems with Universal Face Mask
(Zeljka Zorz)
Pentagon UFO study led by researcher who believes in the supernatural
(Science)
Criminal Justice Algorithm Predicts Risk of Biased Sentencing
(Jule Pattison-Gordon)
The Long, Strange Relationship Between Psychedelics and Telepathy (Vice)
How your brainwaves could be used in criminal trials (techxplore.com)
New 'Retbleed' Speculative Execution Attack Affects AMD, Intel CPUs
(Ravie Lakshmanan)
New UEFI Firmware Vulnerabilities Impact Several Lenovo Notebook
(The Hacker News)
Choosing a non-Windows OS on Lenovo Secured-core PCs is trickier than it
should be (The Register)
How the FBI Wiretapped the World (Vice)
Democracy dies behind a paywall (Poynter)
User Generated Content (Lauren Weinstein)
Cryptomining Capacity in U.S. Rivals Energy Use of Houston (Hiroko Tabuchi)
How the fall of Celsius dragged down crypto investors (CNBC)
Tech experts send letter to Congress urging them to resist crypto industry
lobbying (Twitter)
GM rebate on new Cadillac Lyriq if drivers sign NDA, agree to tracking
(USA Today)
Uber leveraged violent attacks against its drivers to pressure
politicians (WashPost)
About the Uber Files investigation (WashPost)
Hit the kill switch: Uber used covert tech to thwart government raids
(WashPost)
GOOD! - Google bans deepfake-generating AI from Colab (TechCrunch)
Google Voice problems (Lauren Weinstein)
Full text of Google's proposal for political email to bypass Gmail spam
filters -- and an interesting sentence
MIT scientists think they've discovered how to fully reverse climate change
(BGR)
Meet the Lobbyist Next Door (WiReD)
Facebook encrypting links to avoid URL-stripping (Henry Baker)
Facebook, privacy and abortion (Reveal News)
Nobody likes self-checkout. Here's why it's everywhere (The Atlantic)
Major American Companies to Schools: Expand Access to Computer Science
(Alyson Klein)
FedEx bot apologizes for pending delivery' of missing human remains
(WashPost)
Re: Canadian network outage misunderstatement OTD (David W. Hodgins)
ISODARCO 2023 (Diego.Latella)
Abridged info on RISKS (comp.risks)

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

Date: Mon, 18 Jul 2022 19:38:53 -0400
From: Gabe Goldberg <gabe@gabegold.com>
Subject: The Big Hack: How China Used a Tiny Chip to Infiltrate
U.S. Companies (Bloomberg)

The attack by Chinese spies reached almost 30 U.S. companies, including
Amazon and Apple, by compromising America's technology supply chain,
according to extensive interviews with government and corporate sources.

https://www.bloomberg.com/news/features/2018-10-04/the-big-hack-how-china-used-a-tiny-chip-to-infiltrate-america-s-top-companies

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

Date: Mon, 18 Jul 2022 15:31:08 -0400
From: Monty Solomon <monty@roscom.com>
Subject: Driver says GPS made him turn onto train tracks in Everett; at
least he was able to escape before train destroyed his car
(UniversalHub)

https://www.universalhub.com/2022/driver-says-gps-made-him-turn-train-tracks-everett

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

Date: Wed, 13 Jul 2022 11:59:21 -0400 (EDT)
From: ACM TechNews <technews-editor@acm.org>
Subject: DeepMind AI Learns Simple Physics Like a Baby (Davide Castelvecchi)

Davide Castelvecchi, *Nature*, 11 Jul 2022,
via ACM TechNews; 13 Jul 2022

Computer scientists at the DeepMind artificial intelligence (AI) research
laboratory trained a software model to learn simple physical rules about
object behavior. The researchers trained the Physics Learning through
Auto-encoding and Tracking Objects (PLATO) neural network model using
animated videos and images of objects like cubes and balls, in order for it
to generate an internal representation of the physical properties of each
object. The model learned patterns such as continuity, solidity, and
persistence of shape. DeepMind's Luis Piloto said the software makes
predictions at every step in the video, and its accuracy increases as the
video progresses. Piloto suggested PLATO could be a first step toward AI
that can test theories about how human infants learn.

https://orange.hosting.lsoft.com/trk/click?ref=znwrbbrs9_6-2ee75x234badx070806&

[Interesting metaphor. How long dies it take a baby to understand quantum
theory and space physics? Through elementary and secondary schools,
universities, and specialized grad schools? Would you want that baby to
grow into building your airplanes without the benefits of a real in-person
education, or even designing your space ship so that you might some day
want to escape from this planet? PGN]

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

Date: Mon, 18 Jul 2022 12:25:28 -0400 (EDT)
From: ACM TechNews <technews-editor@acm.org>
Subject: As AI Language Skills Grow, So Do Scientists' Concerns
(Matt O'Brien)

Matt O'Brien, Associated Press, 17 Jul 2022
via ACM TechNews; Monday, July 18, 2022

Scientists are worried about the use of large language models in chatbots
and other technologies, not least because their creators conceal their inner
workings and the flaws that can cause such systems to spread misinformation.
Stanford University's Percy Liang said companies face competitive pressure
not to expose large language models' underpinning technology, or to partner
on community standards. A group of scientists worked with France's
government to launch the BigScience Large Open-science Open-access
Multilingual Language Mode (BLOOM) large language model, which was developed
to counter closed models like Microsoft's GPT-3. BLOOM functions across 46
languages, while most systems concentrate on English or Chinese.

https://orange.hosting.lsoft.com/trk/click?ref=znwrbbrs9_6-2eeb3x234c60x070732&

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

Date: Wed, 13 Jul 2022 11:59:21 -0400 (EDT)
From: ACM TechNews <technews-editor@acm.org>
Subject: Researchers Defeat Facial Recognition Systems with Universal Face
Mask (Zeljka Zorz)

Zeljka Zorz, *Help Net Security*, 12 Jul 2022,
via ACM TechNews; 13 Jul 2022

Researchers at Israel's Ben-Gurion University of the Negev (BGU) and Tel
Aviv University found that facial recognition (FR) systems may be thwarted
by fabric face masks boasting adversarial patterns. The researchers employed
a gradient-based optimization process to generate a universal perturbation
and mask to falsely classify each wearer as an unknown identity. BGU's Alon
Zolfi said, "The perturbation depends on the FR model it was used to attack,
which means different patterns will be crafted depending on the different
victim models." Zolfi suggested FR models could see through masked face
images by training them on images containing adversarial patterns, by
teaching them to make predictions based only on the upper area of the face,
or by training them to generate lower facial areas based on upper facial
areas.

https://orange.hosting.lsoft.com/trk/click?ref=znwrbbrs9_6-2ee75x234bacx070806&

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

Date: Mon, 18 Jul 2022 20:09:01 -0400
From: Gabe Goldberg <gabe@gabegold.com>
Subject: Pentagon UFO study led by researcher who believes in the
supernatural (AAAS Science)

Critics dumbfounded by reality TV star Travis Taylor's position as "chief
scientist"

https://www.science.org/content/article/pentagon-ufo-study-led-researcher-who-believes-supernatural

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

Date: Wed, 13 Jul 2022 11:59:21 -0400 (EDT)
From: ACM TechNews <technews-editor@acm.org>
Subject: Criminal Justice Algorithm Predicts Risk of Biased Sentencing
(Jule Pattison-Gordon)

Jule Pattison-Gordon, *Government Technology*, 12 Jul 2022,
via ACM TechNews; 13 Jul 2022

Members of the American Civil Liberties Union, Carnegie Mellon University,
the Idaho Justice Project, and the University of Pennsylvania developed a
criminal justice algorithm to predict the probability of defendants
receiving biased sentences in court. The algorithm factors in seemingly
immaterial variables like the judge's and defendant's gender and race, along
with case details like mandatory minimum sentencing requirements and the
nature of the offense, to forecast how likely the judge is to issue an
unusually long sentence (longer than those issued in 90% of the other cases
with "identical legally relevant factors"). The team of developers suggest
the algorithm could help potentially wronged defendants argue for reducing
disproportionately harsh sentences.