RISKS-LIST: Risks-Forum Digest Tuesday 31 May 2022 Volume 33 : Issue 24 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/33.24> The current issue can also be found at <http://www.csl.sri.com/users/risko/risks.txt> Contents: When a machine invents things for humanity, who gets the patent? (techxplore) Inside the Government Fiasco That Nearly Closed the U.S. Air System (ProPublica) Serious Warning Issued For Millions Of Google Gmail Users (Forbes) 2022 Data Breach Investigations Report (DBIR) Children's Rights Violations by Governments that Endorsed Online Learning During the Covid-19 Pandemic (HRW) Elon Musk: When He saw the Tesla CEO for who he really is. (S;ate) Help Wanted: State Misinformation Sheriff (Jose Maria Mateos) Microsoft Wants to Prove You Exist with Verified ID System, if You'll Let It (Kyle Barr) An Autonomous Car Blocked a Fire Truck Responding to an Emergency (WiReD) Re: Autonomous vehicles can be tricked into dangerous driving (Martin Ward, Richard Stein) Re: Artificial intelligence predicts patients' race from their medical images (Jan Wolitzky, Amos Shapir, Steve Bacher) Security and Human Behaviour 2022 (Jose Maria Mateos) Abridged info on RISKS (comp.risks) ---------------------------------------------------------------------- Date: Sat, 28 May 2022 13:34:10 +0800 From: Richard Stein <rmst...@ieee.org> Subject: When a machine invents things for humanity, who gets the patent? (techxplore.com) https://techxplore.com/news/2022-05-machine-humanity-patent.html "The day is coming -- some say has already arrived -- when artificial intelligence starts to invent things that its human creators could not. But our laws are lagging behind this technology, UNSW experts say. "It's not surprising these days to see new inventions that either incorporate or have benefitted from artificial intelligence (AI) in some way, but what about inventions dreamt up by AI -- do we award a patent to a machine?" The authors argue that a new class of intellectual property, that created or discovered by AI (AI-IP), be established to enable patent rights protection and adjudication. Would an anti-AI-IP invention, a dataset or learning model or combination that can defeat an AI-IP's operation be eligible for patent, or would it be considered dangerous malware? ------------------------------ Date: Sat, 28 May 2022 14:40:07 -0400 From: Gabe Goldberg <g...@gabegold.com> Subject: Inside the Government Fiasco That Nearly Closed the U.S. Air System (ProPublica) The upgrade to 5G was supposed to bring a paradise of speedy wireless. But a chaotic process under the Trump administration, allowed to fester by the Biden administration, turned it into an epic disaster. The problems haven't been solved. The prospect sounded terrifying. A nationwide rollout of new wireless technology was set for January, but the aviation industry was warning it would cause mass calamity: 5G signals over new C-band networks could interfere with aircraft safety equipment, causing jetliners to tumble from the sky or speed off the end of runways. Aviation experts warned of "catastrophic failures leading to multiple fatalities." [...] But the Trump administration didn't initially seem inclined to leave 5G decisions to the FCC. The administration saw the fifth generation of cellular technology, with its faster speeds and automation efficiencies for industry, as its single biggest communications initiative. Top Trump officials viewed the technology through the prism of competition with China. Many in the administration also expressed fears that Huawei Technologies, a dominant maker of 5G hardware, might be a conduit for Chinese government surveillance, posing a national-security threat. (Huawei has always denied such claims.) Trump lieutenants began employing a nationalist battle cry: America needed to "win the race to 5G" against China. https://www.propublica.org/article/fcc-faa-5g-planes-trump-biden ------------------------------ Date: Sat, 21 May 2022 18:17:34 -1000 From: geoff goodfellow <ge...@iconia.com> Subject: Serious Warning Issued For Millions Of Google Gmail Users Gmail is the world's most popular email service, it is also known as one of the most secure. But a dangerous exploit might make you rethink how you want to use the service in future. In an eye-opening *blog post* <https://ysamm.com/?p=763>, security researcher Youssef Sammouda has revealed that Gmail's OAuth authentication code enabled him to exploit vulnerabilities in Facebook to hijack Facebook accounts when Gmail credentials are used to sign in to the service. And the wider implications of this are significant. Speaking to *The Daily Swing* <https://portswigger.net/daily-swig/facebook-account-takeover-researcher-scoops-40k-bug-bounty-for-chained-exploit>, Sammouda explained that he was able to exploit redirects in Google OAuth and chain it with elements of Facebook's logout, checkpoint and sandbox systems to break into accounts. Google OAuth is part of the '*Open Authorization* <https://en.wikipedia.org/wiki/OAuth>' standard used by Amazon, Microsoft, Twitter and others which allows users to link accounts to third-party sites by signing into them with the existing usernames and passwords they have already registered with these tech giants. Sammouda reports no vulnerabilities using other email accounts. He does stress that it could potentially be applied more widely "but that was more complicated to develop an exploit for." He states Facebook paid him a $44,625 'bug bounty' for its role in this vulnerability. Facebook has subsequently patched the vulnerability from their side. I have contacted Google for a response on the role of Google OAuth in the exploit and will update this post when/if I receive a reply. Commenting on Sammouda's findings, security provider *Malwarebytes Labs* <https://blog.malwarebytes.com/exploits-and-vulnerabilities/2022/05/gmail-linked-facebook-accounts-vulnerable-to-attack-using-a-chain-of-bugs-now-fixed/> issued a warning to anyone using linked accounts: "Linked accounts were invented to make logging in easier," writes Pieter Arntz, the company's Malware Intelligence Researcher. "You can use one account to log in to other apps, sites and services... All you need to do to access the account is confirm that the account is yours." [...] https://www.forbes.com/sites/gordonkelly/2022/05/21/google-gmail-security-facebook-oauth-login-warning/ ------------------------------ Date: Sun, 29 May 2022 11:45:20 -0400 From: Monty Solomon <mo...@roscom.com> Subject: 2022 Data Breach Investigations Report (DBIR) https://www.verizon.com/business/resources/reports/dbir/ https://www.verizon.com/business/resources/reports/2022/dbir/2022-dbir-data-breach-investigations-report.pdf Verizon DBIR: Stolen credentials led to nearly 50% of attacks The 2022 Verizon Data Breach Investigations Report revealed enterprises' ongoing struggle with securing credentials and avoiding common mistakes such as misconfigurations. https://www.techtarget.com/searchsecurity/news/252520686/Verizon-DBIR-Stolen-credentials-led-to-nearly-50-of-attacks ------------------------------ Date: Sun, 29 May 2022 14:48:34 -0400 From: Gabe Goldberg <g...@gabegold.com> Subject: Children's Rights Violations by Governments that Endorsed Online Learning During the Covid-19 Pandemic (HRW) How Dare They Peep into My Private Life? This report is a global investigation of the education technology (EdTech) endorsed by 49 governments for children's education during the pandemic. Based on technical and policy analysis of 164 EdTech products, Human Rights Watch finds that governments' endorsements of the majority of these online learning platforms put at risk or directly violated children's privacy and other children's rights, for purposes unrelated to their education. The coronavirus pandemic upended the lives and learning of children around the world. Most countries pivoted to some form of online learning, replacing physical classrooms with EdTech websites and apps; this helped fill urgent gaps in delivering some form of education to many children. But in their rush to connect children to virtual classrooms, few governments checked whether the EdTech they were rapidly endorsing or procuring for schools were safe for children. As a result, children whose families were able to afford access to the Internet and connected devices, or who made hard sacrifices in order to do so, were exposed to the privacy practices of the EdTech products they were told or required to use during Covid-19 school closures. https://www.hrw.org/report/2022/05/25/how-dare-they-peep-my-private-life/childrens-rights-violations-governments ------------------------------ Date: Mon, 30 May 2022 16:22:31 -0400 From: Gabe Goldberg <g...@gabegold.com> Subject: Elon Musk: When He saw the Tesla CEO for who he really is. The CEO's mythmaking often obscures an uglier truth. The public is finally reckoning with it. Edward Niedermeyer: This duplicity on Tesla's part, I reasoned, couldn't be a mere accident. To borrow the folksy saying favored by Warren Buffett: There is never just one cockroach. So I began digging into every aspect of Tesla's business, and in the years that followed, my investigations turned up no shortage of cockroaches. https://slate.com/technology/2022/05/elon-musk-tesla-twitter-fables.html ------------------------------ Date: Tue, 31 May 2022 06:05:21 -0400 From: =?iso-8859-1?Q?Jos=E9_Mar=EDa?= Mateos <ch...@rinzewind.org> Subject: Help Wanted: State Misinformation Sheriff https://www.nytimes.com/2022/05/31/technology/misinformation-sheriff-election-midterms.html > Ahead of the 2020 elections, Connecticut confronted a bevy of falsehoods > about voting that swirled around online. One, widely viewed on Facebook, > wrongly said that absentee ballots had been sent to dead people. On > Twitter, users spread a false post that a tractor-trailer carrying ballots > had crashed on Interstate 95, sending thousands of voter slips into the > air and across the highway. > Concerned about a similar deluge of unfounded rumors and lies around this > year's midterm elections, the state plans to spend nearly $2 million on > marketing to share factual information about voting, and to create its > first-ever position for an expert in combating misinformation. With a > salary of $150,000, the person is expected to comb fringe sites like > 4chan, far-right social networks like Gettr and Rumble and mainstream > social media sites to root out early misinformation narratives about > voting before they go viral, and then urge the companies to remove or flag > the posts that contain false information. "What do you do for a living?" "I... er... browse 4chan" ------------------------------ Date: Tue, 31 May 2022 11:25:10 -0700 From: Lauren Weinstein <lau...@vortex.com> Subject: Microsoft Wants to Prove You Exist with Verified ID System, if You'll Let It (Kyle Barr) Kyle Barr, Gizmodo, 31 May 2022 In the decade-spanning conflict between the need for online privacy and efforts to stop fake accounts from accessing sensitive info, the tech monolith that is Microsoft is putting its massive weight behind the creation of standardized online identities. In its announcement Tuesday, Microsoft talked up its Entra management systems that includes Verified ID, promoting itself as a quick way of giving sensitive identification to entities that need to verify that you are who you say you are. In its release, the company said that old means of restricting electronic access was "no longer sustainable" because of how digital estates have become "boundary-less." What that really means is people abusing fake accounts to gain access to sensitive online networks have created a host of issues for private companies, governments and more. Microsoft itself has been targeted by hackers who managed to access company information on Microsoft's Azure cloud computing platform. The LAPSUS$ group of hackers has previously called on tech company employees to give them sensitive info. In at least a few of these cases, hackers were able to gain access to sensitive networks by using stolen account details to log in. Recently, reports showed hackers were able to gain user data from tech companies by posing as law enforcement officials. Instead of having personal information spread across a host of apps and services, this Verified ID system acts as a kind of digital wallet or personal info portfolio that can be handed over to employers, bankers, or whoever needs a verified identification. Ankur Patel, Microsoft's principal programmer for digital identity, told Protocol the new system could include college diplomas, bank notes, or even doctors' notes for a clean bill of health. Those who create and issue verified IDs can also suspend or invalidate credentials after they're issued. https://gizmodo.com/microsoft-verified-id-entra-digital-identity-wallet-1848996341 [ID is a very complex problem. But I will not be an early adopter of this. -L] ------------------------------ Date: Fri, 27 May 2022 19:51:30 -0400 From: Gabe Goldberg <g...@gabegold.com> Subject: An Autonomous Car Blocked a Fire Truck Responding to an Emergency (WiReD) The incident in San Francisco cost first responders valuable time, and underscores the challenges Cruise and other companies face in launching driverless taxis. https://www.wired.com/story/cruise-fire-truck-block-san-francisco-autonomous-vehicles/ It's the old "free will vs. determinism" debate, and determinism lost. Pretty soon my toaster will say, "I'm sorry Gabe, I can't do that" when I tell it how I like my toast. ------------------------------ Date: Sat, 28 May 2022 16:47:51 +0100 From: Martin Ward <mar...@gkc.org.uk> Subject: Re: Autonomous vehicles can be tricked into dangerous driving behavior (RISKS-33.23) > Without human-like, contextual interpretation and reasoning, an AV's CAS > cannot discriminate a cardboard box from a concrete block. What if there is a cardboard box covering a concrete block? [RISKS readers would like to believe that every cardboard box should be avoided, because it might house a homeless person or your favorite pet. PGN] ------------------------------ Date: Sun, 29 May 2022 08:50:25 +0800 From: Richard Stein <rmst...@ieee.org> Subject: Re: Autonomous vehicles can be tricked into dangerous driving behavior (Ward, RISKS-33.24) > What if there is a cardboard box covering a concrete block? People are required to purchase auto and health insurance policies. A cardboard box covering a concrete block which instantaneously confronts a vehicle steered by either human or machine are visually indistinguishable. A detected radio-wave (or infrared) signature would reveal a different signature for the machine to reconcile against the visual. An obstacle encountered under these circumstances favors neither human nor machine, unless the machine is train-sized or bigger. ------------------------------ Date: Fri, 27 May 2022 20:06:30 -0400 From: Jan Wolitzky <jan.wolit...@gmail.com> Subject: Re: Artificial intelligence predicts patients' race from their medical images (RISKS-33.23) Care should be taken in interpreting the results of a study that purports to use objective data and artificial intelligence to predict an objectively undefined variable such as race, a social construct. The "race" of the subjects in the training set was entirely subjective, i.e., self-reported. The authors never specify, e.g., how many different racial categories the subjects in each dataset were allowed to choose from, or whether there were differences in this among the datasets. Furthermore, the datasets used were predominantly from an institution in Georgia, a historically racist area in a historically racist country, so the objective value of "race" assigned must be taken with a very large grain of salt. ------------------------------ From: Amos Shapir <amos...@gmail.com> Date: Sat, 28 May 2022 12:51:30 +0300 Subject: Re: Artificial intelligence predicts patients' race from their medical images (medicalxpress, RISKS-33.23) Ethnic identity is in the eye of the beholder. These AI medical systems are not different in principle from any other medical examination; the only parameters they can detect are physical properties of a patient's body. It's true that such properties (most notably, skin color) have been, and still are, used for discrimination against people, but that should not affect the technicalities of medical procedures. Mixing these with social and political issues might end in, e.g., labeling tests for sickle-cell anemia or vitamin D deficiency as racist. ------------------------------ Date: Sat, 28 May 2022 17:32:30 +0000 (UTC) From: Steve Bacher <seb...@verizon.net> Subject: Re: Artificial intelligence predicts patients' race from their medical images (medicalxpress, RISKS-33.23) Yes, but it could also be used to positive ends, like identifying patients prone to sickle cell anemia, for instance. Or it could be the basis for corrective, reparative, or anti-profiling policies, whether you agree with those or not. After all, it's just information, and can be put to good or bad uses. [One of the lessons from RISKS is that many things are dual-use -- good or bad. This is just one more. Discriminating between them might be like trying to use technology to mediate the fairness of an ill-conceived *duel*, which would also have *dual* uses, especially if the technology could be easily rigged, like so many other things. PGN] ------------------------------ Date: Tue, 31 May 2022 06:06:13 -0400 From: =?iso-8859-1?Q?Jos=E9_Mar=EDa?= Mateos <ch...@rinzewind.org> Subject: Security and Human Behaviour 2022 Seen on Bruce Schneier's blog: https://www.cl.cam.ac.uk/~rja14/shb22/. This is the list of working papers for the conference, which I think will be of interest to many RISKS subscribers. ------------------------------ Date: Mon, 1 Aug 2020 11:11:11 -0800 From: risks-requ...@csl.sri.com Subject: Abridged info on RISKS (comp.risks) The ACM RISKS Forum is a MODERATED digest. Its Usenet manifestation is comp.risks, the feed for which is donated by panix.com as of June 2011. => SUBSCRIPTIONS: The mailman Web interface can be used directly to subscribe and unsubscribe: http://mls.csl.sri.com/mailman/listinfo/risks => SUBMISSIONS: to risks@CSL.sri.com with meaningful SUBJECT: line that includes the string `notsp'. Otherwise your message may not be read. *** This attention-string has never changed, but might if spammers use it. => SPAM challenge-responses will not be honored. Instead, use an alternative address from which you never send mail where the address becomes public! => The complete INFO file (submissions, default disclaimers, archive sites, copyright policy, etc.) is online. <http://www.CSL.sri.com/risksinfo.html> *** Contributors are assumed to have read the full info file for guidelines! => OFFICIAL ARCHIVES: http://www.risks.org takes you to Lindsay Marshall's searchable html archive at newcastle: http://catless.ncl.ac.uk/Risks/VL.IS --> VoLume, ISsue. Also, ftp://ftp.sri.com/risks for the current volume/previous directories or ftp://ftp.sri.com/VL/risks-VL.IS for previous VoLume If none of those work for you, the most recent issue is always at http://www.csl.sri.com/users/risko/risks.txt, and index at /risks-33.00 ALTERNATIVE ARCHIVES: http://seclists.org/risks/ (only since mid-2001) *** NOTE: If a cited URL fails, we do not try to update them. Try browsing on the keywords in the subject line or cited article leads. Apologies for what Office365 and SafeLinks may have done to URLs. ==> Special Offer to Join ACM for readers of the ACM RISKS Forum: <http://www.acm.org/joinacm1> ------------------------------ End of RISKS-FORUM Digest 33.24 ************************