Please try the URL privacy information feature enabled by clicking the flashlight icon above. This will reveal two icons after each link the body of the digest. The shield takes you to a breakdown of Terms of Service for the site - however only a small number of sites are covered at the moment. The flashlight take you to an analysis of the various trackers etc. that the linked site delivers. Please let the website maintainer know if you find this useful or not. As a RISKS reader, you will probably not be surprised by what is revealed…
Tractors have ground to a halt in paddocks across Australia and New Zealand because of a signal failure in the satellite farmers use to guide their GPS-enabled machinery, stopping them from planting their winter crop. The satellite failure on Monday was a bolt from the blue for farmers in NSW and Victoria, who were busy taking advantage of optimal planting conditions for crops including wheat, canola, oats, barley and legumes. “You couldn't have picked a worse time for it,''D said Justin Everitt, a grain grower in the Riverina who heads NSW Farmers' grains committee. “Over the past few years, all these challenges have been thrown at us, but this is just one we never thought would come up.'' Tractors that pull seed-planting machinery, as well as the massive combine harvesters that reap Australia's vast grain crops, are high-tech beasts that can cost hundreds of thousands of dollars. They are enabled with GPS tracking and can be guided to an accuracy within two centimetres, enabling seed-planting equipment to sow crops with precision to drive up efficiency, prevent wastage and boost environmental sustainability. All that went out the window when the Inmarsat-41 satellite signal failed. Katie McRobert, general manager at the Australia Farm Institute, said Australian farmers sourced their GPS signal from one satellite, which was a critical risk to rural industries. Having all your GPS eggs in one basket is a vulnerability on a good day, and a fatal weakness on a bad one,'' McRobert said. “If the Medibank and Optus data breaches didn't make the agriculture industry sit up and take notice, the implementation of kill switches on stolen Ukrainian tractors in 2022 should have been a three-alarm wake-up call. [...] https://www.smh.com.au/national/farmers-crippled-by-satellite-failure-as-gps-guided-tractors-grind-to-a-halt-20230418-p5d1de.html
Bernie Cosell asked Victor Miller a question, which Victor referred to me. This is very strange: this morning, my cell phone thinks it is August 18th 2003. It is *supposed* to get the time/date from the network. What could have caused this? I guess I can turn the network off and put in the right time/date by hand, but any ideas how my phone could have gotten so confused?? Apparently it's just one more 1024-week turnover, as reported in RISKS-20.07 The reset is apparently receiver-dependent, e.g., resetting to 6 Jan 1980 or the previous reset date, as in Bernie's case: THE POTENTIAL RESETTING OF GLOBAL POSITIONING SYSTEM (GPS) RECEIVER INTERNAL CLOCKS 1 Introduction 1.1 The timing mechanism within GPS satellites may cause some GPS equipment to cease to function after 22 August 1999 due to a coding problem. The GPS measures time in weekly blocks of seconds starting from 6 January 1980. For example, at midday on Tuesday 17 September 1996, the system indicates week 868 and 302,400 seconds. However, the software in the satellites' clocks has been configured to deal with 1024 weeks. Consequently on 22 August 1999 (which is week 1025, some GPS receivers may revert to week one (i.e., 6 January 1980). 1.2 Most airborne GPS equipment manufacturers are aware of the potential problem and either have addressed the problem previously, or are working to resolve it. However, there may be some GPS equipment (including portable and hand held types) currently used in aviation that will be affected by this potential problem. 2 Action to be taken by Aircraft Operators Aircraft operators, who use GPS equipment (including portable and hand held types), as additional radio equipment to the approved means of navigation, should enquire from the GPS manufacturer whether the GPS equipment will exhibit the problem. Equipment that exhibits the problem must not be used after 21 August 1999 and either be removed from the aircraft or its operation inhibited. For the Civil Aviation Authority, Safety Regulation Group, Aviation House, Gatwick Airport South, West Sussex RH6 OYR Does anyone know if there have been any desire to automagically fix this problem? or do we just continue to kick the can down another 1024 days? PGN
And this is no April Fool's joke, is it. All of the articles from " In Gen Z's world of dupes, fake is fabulous—until you try it on <https://catless.ncl.ac.uk/Risks/33/68#subj2>" through " AI-Powered Vehicle Descriptions: Save Money, Save, Time, Sell More! <https://catless.ncl.ac.uk/Risks/33/68#subj9>" are missing. The first article ends with a link from the ninth article, which was strange in itself. [I don't think I have ever had an emacs moment like this, Where I managed to lose a large chunk of something without immediately noticing it and being able to yank the deleted text back—in this case *after* the complete issue had been spelling checked and date checked, all set up with the final insertion of the grep-generated ToC in the right order. Perhaps I tried incorrectly to move one item to a different position in the issue. If anyone has a pargticular hankering for the missing items, you might try browsing on the Subject: line of the missing item. I am very short of spare time at the moment, and seriously backlogged since 7 April. I also may have lost a few items from the week before 1 April in the shuffle. However, now it feels like water under the bridge. Here's a start. Bummer. PGN]
As car owners grow hip to one form of theft, crooks are turning to new ones. When a London man discovered the front left-side bumper of his Toyota RAV4 torn off and the headlight partially dismantled not once but twice in three months last year, he suspected the acts were senseless vandalism. When the vehicle went missing a few days after the second incident, and a neighbor found their Toyota Land Cruiser gone shortly afterward, he discovered they were part of a new and sophisticated technique for performing keyless thefts. It just so happened that the owner, Ian Tabor, is a cybersecurity researcher specializing in automobiles. While investigating how his RAV4 was taken, he stumbled on a new technique called CAN injection attacks. https://arstechnica.com/information-technology/2023/04/crooks-are-stealing-cars-using-previously-unknown-keyless-can-injection-attacks/ [Hacking the CAN-bus is hardly new, but this attack seems relatively easy. PGN]
Artificial intelligence has advanced quite a bit in recent years, so much so that it is now equipped to reject humans. A 40-year-old man in California recently confessed that he fell in love with an AI chatbot but was heartbroken when she rejected his "steamy" advances. https://www.indiatimes.com/trending/wtf/divorced-man-falls-in-love-with-ai-chatbot-phaedra-598271.html
Idan Mor is an Israeli comic actor, who had invented and played a stage character named Gadi Wilcherski. He became famous when he had taken to participating in anti-government demonstrations in character, and even been interviewed on main channels, initially without them being aware that the person they were . The Wilcherski character had gotten a life of his own, being politically active, even appearing before Knesset committees, and of course had his own Facebook page. But last week, after the actor Idan Mor had uploaded to his FB page images of himself playing his stage persona, FB (as usual) had acted as judge, jury and executioner, and closed down Mor's page on the pretext of "impersonating a real living person". It seems that this "hall of mirrors" situation was just too much for FB's fact checking.
The latest `intelligence leak' that is all over the media is, quite likely, yet another Manning/Snowden thing. But, given a lot of the details that are starting to come out, I can't help but thinking that it could very well be another type of discord attack ...
Win, Lose, or Draw? Making Sense of Settlement *The New York Times*, Business section, 20 Apr 2023, National edition, pages B1, B4, B5 (three-page spread) 1. The Victor: Tiffany Hsu Dominion emerges in a stronger position to win back sksittish clients and score new business. 2. The First Amendment: David Enrich Assaults are likely to continue on a a landmark Supreme Court ruling that protects the media, 3. The Reaction: Michael M. Grynbaum Some critics of Fox News were hoping Murdoch and the network would face a stiffer penalty. 4. Critic's Notebook: James Poniewozik The network's main goal became the maintenance of a reality bubble that its hosts helped shape.
Early adopters thought cryptocurrencies would be free from prying eyes. But tracking the flow of funds has become a big business. https://www.nytimes.com/2023/04/22/business/crypto-blockchain-tracking-chainalysis.html
*Bulletin of Atomic Scientists* https://thebulletin.org/2023/04/to-avoid-an-ai-arms-race-the-world-needs-to-expand-scientific-collaboration/#post-heading
The AI chatbot falsely told users that Australia's Hepburn Shire mayor Brian Hood was jailed in an international bribery scandal. He was actually the *whistleblower*. He may sue. https://www.washingtonpost.com/technology/2023/04/06/chatgpt-australia-mayor= -lawsuit-lies/ [Another quirky risk of whistle-blowing. PGN]
https://www.cbc.ca/news/world/openai-chatgpt-data-privacy-investigations-1.6804205 As governments rush to address concerns about the rapidly-advancing generative artificial intelligence industry, experts in the field say greater oversight is needed over what data is used to train the systems. Earlier this month, Italy's data protection agency launched a probe of OpenAI <https://www.cbc.ca/news/world/italy-openai-chatgpt-ban-1.6797963> and temporarily banned ChatGPT, their AI-powered chatbot. On Tuesday, Canada's privacy commissioner also announced an investigation of OpenAI <https://www.cbc.ca/news/politics/privacy-commissioner-investigation-openai-chatgpt-1.6801296>. Both agencies cited concerns around data privacy.
Last month one of our journalists received an interesting email. A researcher had come across mention of a Guardian article, written by the journalist on a specific subject from a few years before. But the piece was proving elusive on our website and in search. Had the headline perhaps been changed since it was launched? Had it been removed intentionally from the website because of a problem we'd identified? Or had we been forced= to take it down by the subject of the piece through legal means? The reporter couldn't remember writing the specific piece, but the headline certainly sounded like something they would have written. It was a subject they were identified with and had a record of covering. Worried that there may have been some mistake at our end, they asked colleagues to go back through our systems to track it down. Despite the detailed records we keep of all our content, and especially around deletions or legal issues, they could find no trace of its existence. Why? Because it had never been written. Luckily the researcher had told us that they had carried out their research using ChatGPT. In response to being asked about articles on this subject, the AI had simply made some up. Its fluency, and the vast training data it is built on, meant that the existence of the invented piece even seemed believable to the person who absolutely hadn't written it. Huge amounts have been written about generative AI's tendency to manufacture facts and events. But this specific wrinkle—the invention of sources -- is particularly troubling for trusted news organisations and journalists whose inclusion adds legitimacy and weight to a persuasively written fantasy. And for readers and the wider information ecosystem, it opens up whole new questions about whether citations can be trusted in any way, and could well feed conspiracy theories about the mysterious removal of articles on sensitive issues that never existed in the first place. [...] https://www.theguardian.com/commentisfree/2023/apr/06/ai-chatgpt-guardian-technology-risks-fake-article
Professors are using ChatGPT detector tools to accuse students of cheating. But what if the software is wrong? <#> Universities, professors and students are grappling with the repercussions of using AI cheating detectors from companies like Turnitin and GPTZero. https://www.usatoday.com/story/news/education/2023/04/12/how-ai-detection-tool-spawned-false-cheating-case-uc-davis/11600777002/ [My solution to teaching about Chat GPT is not to ask for essays, but rather to provide each student with a crafted essay of specific relevance to the course and ask the class to write a thorough analysis of the falsehoods. The thoroughness of the analysis would be a strong indicator of each student's abilities. Team efforts could even be considered in some topics, although the teams should not be the same from topics to topic. This strategy would work for many different types of classses -- history, literature, science, etc. PGN]
https://arxiv.org/abs/2209.15 SoK: On the Impossible Security of Very Large Foundation Models El-Mahdi El-Mhamdi, Sadegh Farhadkhani, Rachid Guerraoui, Nirupam Gupta, L=C3=AA-Nguy=C3=AAn Hoang, Rafael Pinot, John Stephan Large machine learning models, or so-called foundation models, aim to serve as base-models for application-oriented machine learning. Although these models showcase impressive performance, they have been empirically found to pose serious security and privacy issues. We may however wonder if this is a limitation of the current models, or if these issues stem from a fundamental intrinsic impossibility of the foundation model learning problem itself. This paper aims to systematize our knowledge supporting the latter. More precisely, we identify several key features of today's foundation model learning problem which, given the current understanding in adversarial machine learning, suggest incompatibility of high accuracy with both security and privacy. We begin by observing that high accuracy seems to require (1) very high-dimensional models and (2) huge amounts of data that can only be procured through user-generated datasets. Moreover, such data is fundamentally heterogeneous, as users generally have very specific (easily identifiable) data-generating habits. More importantly, users' data is filled with highly sensitive information, and may be heavily polluted by fake users. We then survey lower bounds on accuracy in privacy-preserving and Byzantine-resilient heterogeneous learning that, we argue, constitute a compelling case against the possibility of designing a secure and privacy-preserving high-accuracy foundation model. We further stress that our analysis also applies to other high-stake machine learning applications, including content recommendation. We conclude by calling for measures to prioritize security and privacy, and to slow down the race for ever larger models. It was mentioned in passing in this article: Google's Rush to Win in AI Led to Ethical Lapses, Employees Say The search giant is making compromises on misinformation and other harms in order to catch up with ChatGPT, workers say Davey Alba and Julia Love, Bloomberg, 19 Apr 2023 https://www.bloomberg.com/news/features/2023-04-19/google-bard-ai-chatbot-raises-ethical-concerns-from-employees
And the Grammy goes to an AI music generator? The breakout hit of the spring is *Heart on My Sleeve*—a track that sounds just like a collaboration between musicians Drake and The Weeknd, two mega-popular artists who didn't record any of the song themselves. The track relies purely on AI-generated imitations of their voices, posted online by a pseudonymous TikTok user, and since it went viral last weekend it's been heard millions of times. It has also generated takedown notices and a statement from The Weeknd's label, Universal Music Group, slamming AI-powered copyright infringement and calling for users of the technology to get on the right *side of history*. [long item truncated for RISKS]
Technology companies were once leery of what some artificial intelligence could do. Now the priority is winning control of the industry's next big thing. https://www.nytimes.com/2023/04/07/technology/ai-chatbots-google-microsoft.html
It's hard to believe, but this ad was AI generated. It's not real. The future is here. [...] https://twitter.com/0xgaut/status/1650867275103174660
Preeminent scientists are sounding the alarm that ideology is undermining merit in the sciences. I strongly support them.*—Richard Dawkins *In Defense of Merit in Science Merit is a central pillar of liberal epistemology, humanism, and democracy. The scientific enterprise, built on merit, has proven effective in generating scientific and technological advances, reducing suffering, narrowing social gaps, and improving the quality of life globally. This perspective documents the ongoing attempts to undermine the core principles of liberal epistemology and to replace merit with non-scientific, politically motivated criteria. We explain the philosophical origins of this conflict, document the intrusion of ideology into our scientific institutions, discuss the perils of abandoning merit, and offer an alternative, human-centered approach to address existing social inequalities. *Keywords: STEM; Enlightenment; meritocracy; critical social justice; postmodernism; identity politics; Mertonian norms*... https://journalofcontroversialideas.org/article/3/1/236via https://twitter.com/RichardDawkins/status/1651970327902138370
According to an agency disciplinary database that WIRED obtained through a public records request, ICE investigators found that the organization's agents likely queried sensitive databases on behalf of their friends and neighbors. They have been investigated for looking up information about ex-lovers and coworkers and have shared their login credentials with family members. In some cases, ICE found its agents leveraging confidential information to commit fraud or pass privileged information to criminals for money. https://www.wired.com/story/ice-agent-database-abuse-records/
https://9to5mac.com/2023/04/07/security-breaches-covered-up/ [How do you know? Paradox: If they were any better at covering it up. the reported percent might be smaller! PGN]
Prompt injection— a new type of attack https://www.theregister.com/2023/04/26/simon_willison_prompt_injection/
The bicameral and bipartisan bill, Block Nuclear Launch by Autonomous AI Act of 2023, primarily seeks to mandate a human element in all AI systmes and protocols that govern U.S. nuclear devices. Hope AI does not exceed the president's authority to deploy a nuclear weapon, as satirized by Stanley Kubrick in Dr. Strangelove. https://www.nextgov.com/emerging-tech/2023/04/lawmakers-initiate-several-efforts-put-guardrails-ai-use/385711/
Chinese spy balloon gathered intelligence from sensitive U.S. military sites, despite U.S. efforts to block it The intelligence China collected was mostly from electronic signals, which can be picked up from weapons systems or include communications from base personnel. https://www.nbcnews.com/politics/national-security/china-spy-balloon-collected-intelligence-us-military-bases-rcna77155
In the story linked to by https://www.washingtonpost.com/transportation/2023/03/24/metrorail-ato-train-operator/, it says that the fatal accident that triggered the disablement of ATO was not the failure of the ATO technology itself but of physical components that stopped working. First of all, aren't the physical components to be considered a part of the ATO system, thus meaning that ATO has effectively failed? More important, this is an inevitable eventuality for self-driving cars and related technologies. What happens when the detection devices have mechanical breakdowns? Hopefully the systems are designed with safe failover procedures, at minimum turning control automatically over to the human operator or to a human remote controller if there is no human in the vehicle. But what if they aren't? A mechanical failure of a robot waiter might have relatively minor consequences. But what of a robotic hospital transport system, say? Hopefully we carbon-based units will still be around to repair stuff.
OpenSSL is the hammer for just about every screw related to certificates and encryption and has recently even added mainstream support for key derivation functions (KDF). This class of functions allows for stretching a potentially weak memorized secret into a more resistant authenticator in a systematic manner. OpenSSL has been using passwords and passphrases for a long time for protecting private keys, so there is a whole class of functions for use of those secrets and even some guidance provided for them. https://www.openssl.org/docs/manmaster/man1/openssl-passphrase-options.html "If no password argument is given and a password is required then the user is prompted to enter one. The actual password is`password;. Since the password is visible to utilities (like 'ps' under Unix) this form should only be used where security is not important." So far, so good. You can put the password in the command line, but it is flagged appropriately and alternatives exist that shuffle the password from memory to memory without being exposed to the process list. Not so with openssl's kdf module. https://www.openssl.org/docs/manmaster/man1/openssl-kdf.html "Specifies the password as an alphanumeric string (use if the password contains printable characters only). The password must be specified for PBKDF2 and scrypt." The password isn't a first-order option here, it's only a pass through option. So while this at first glance appears to follow the same format as the passphrase system (pass:secret), "pass" is load bearing. It triggers the execution path for the secret. It cannot be replaced by "env" or "file" to specify alternatives, and there is no default path to "if you didn't specify a secret, it defaults to prompting". The alternatives for other secrets in kdf also have load bearing option flags. You cannot pass the secret as hex except in the clear. Oddly, this module supports the first-order options for "digest", "cipher" and "mac", but somehow missed "pass". penSSL KDF works well, but because there is no secure path to use it, it mind as well not exist.
Please report problems with the web pages to the maintainer