pargentum

Posted by EditorDavid

https://news.slashdot.org/story/26/04/04/2055256/americas-cia-recruited-irans-nuclear-scientists---by-threatening-to-kill-them?utm_source=rss1.0mainlinkanon&utm_medium=feed

A former U.S. spy spoke to The New Yorker about "years of clandestine work for the C.I.A. — which, he said, had 'prevented Iran from getting a nuke'." [Kevin] Chalker told me that, as he understood it, the Pentagon had suggested running commando operations to kill key Iranian scientists, as Israel subsequently did. But the C.I.A. proposed recruiting those scientists to defect, as U.S. spies had once courted Soviet physicists. Chalker paraphrased the agency's pitch: "We can debrief them and learn so much more — and, if they say no, then you can kill them." (A more senior agency official confirmed the broad strokes of his account.) The White House liked the agency's idea, and [president George W.] Bush authorized the C.I.A. to conduct clandestine operations to stop Iran from building a bomb. The C.I.A. program that Chalker described to me became publicly known in 2007, when the Los Angeles Times reported on the existence of an agency project called Brain Drain. But the details of the "invitations" to Iranian scientists have not previously been reported... Chalker typically had about ten minutes to explain, as gently as possible, that he was from the C.I.A., that he had the power to secure the scientist and his family a comfortable new life in the U.S. — and that, if the offer was rejected, the scientist, regrettably, would be assassinated. (Chalker tried to emphasize the happier potential outcome.) Killing a civilian scientist would violate international law. The American government has denied ever doing it, and I found no evidence that the U.S. has carried out any such murders. A former senior agency official familiar with the Brain Drain project told me all that mattered was that Iranian scientists had believed they would be killed, regardless of whether the U.S. actually made good on the threat. And Israel had been conducting a campaign to assassinate Iranian scientists, which made the prospect of lethal reprisal highly plausible. Other former officials with knowledge of the project told me that the C.I.A. sometimes shared intelligence with Mossad which enabled its operatives to locate and kill a scientist. Such information exchanges were kept vague enough to preserve deniability if a more legalistic U.S. Administration later took office... [Chalker] is confident that those who rebuffed him were, in fact, killed — one way or another... One of Chalker's colleagues told me that, against the backdrop of so many Israeli assassinations, Chalker's interactions with Iranian scientists could almost be considered humanitarian — he had been "throwing them a lifeline." Of the many scientists he approached, three-quarters ultimately agreed to coöperate. Their 10,000-word article suggests Chalker may now be resentful the CIA didn't help him in a later unrelated lawsuit, noting it's "nearly unheard of for ex-spies to divulge their past activities." But Chalker also says he "helped obtain pivotal information that laid the groundwork for more than a decade of American efforts to disrupt the Iranian nuclear-weapons program, from the Stuxnet cyberattacks, which occurred around 2010 [destroying 1,000 uranium-enriching centrifuges], to the Obama Administration's nuclear deal, in 2015, to the U.S. air strikes on Iranian atomic-energy facilities in the summer of 2025."

Read more of this story at Slashdot.

https://news.slashdot.org/story/26/04/04/2055256/americas-cia-recruited-irans-nuclear-scientists---by-threatening-to-kill-them?utm_source=rss1.0mainlinkanon&utm_medium=feed

Posted by EditorDavid

https://slashdot.org/story/26/04/04/1919236/anthropic-announces-claude-subscribers-must-now-pay-extra-to-use-openclaw?utm_source=rss1.0mainlinkanon&utm_medium=feed

Anthropic's making a big and sudden change — and connecting its Claude AI to third-party agentic tools "is about to get a lot more expensive," writes the Verge: Beginning April 4th at 3PM ET, users will "no longer be able to use your Claude subscription limits for third-party harnesses including OpenClaw," according to an email sent to users on Friday evening. Instead, if users want to use OpenClaw with Claude, they'll have to use a "pay-as-you-go option" that will be billed separate from their Claude subscription. Anthropic's announcement added these extra usage bundles are "now available at a discount." Users can also try Anthropic's API, notes VentureBeat, "which charges for every token of usage rather than allowing for open-ended usage up to certain limits, as the Pro and Max plans have allowed so far. " The technical reality, according to Anthropic, is that its first-party tools like Claude Code, its AI vibe coding harness, and Claude Cowork, its business app interfacing and control tool, are built to maximize "prompt cache hit rates" — reusing previously processed text to save on compute. Third-party harnesses like OpenClaw often bypass these efficiencies... [Claude Code creator Boris Cherny explained on X that "I did put up a few PRs to improve prompt cache hit rate for OpenClaw in particular, which should help for folks using it with Claude via API/overages."] Growth marketer Aakash Gupta observed on X that the "all-you-can-eat buffet just closed," noting that a single OpenClaw agent running for one day could burn $1,000 to $5,000 in API costs. "Anthropic was eating that difference on every user who routed through a third-party harness," Gupta wrote. "That's the pace of a company watching its margin evaporate in real time." However, Peter Steinberger, the creator of OpenClaw who was recently hired by OpenAI, took a more skeptical view of the "capacity" argument."Funny how timings match up," Steinberger posted on X. "First they copy some popular features into their closed harness, then they lock out open source." Indeed, Anthropic recently added some of the same capabilities that helped OpenClaw catch-on — such as the ability to message agents through external services like Discord and Telegram — to Claude Code... User @ashen_one, founder of Telaga Charity, voiced a concern likely shared by other small-scale builders: "If I switch both [OpenClaw instances] to an API key or the extra usage you're recommending here, it's going to be far too expensive to make it worth using. I'll probably have to switch over to a different model at this point." "I know it sucks," Cherny replied. "Fundamentally engineering is about tradeoffs, and one of the things we do to serve a lot of customers is optimize the way subscriptions work to serve as many people as possible with the best mode..." OpenAI appears to be positioning itself as a more "harness-friendly" alternative, potentially using this moment as a customer acquisition channel for disgruntled Claude power users. By restricting subscription limits to their own "closed harness," Anthropic is asserting control over the UI/UX layer. This allows them to collect telemetry and manage rate limits more granularly, but it risks alienating the power-user community that built the "agentic" ecosystem in the first place. Anthropic's decision is a cold calculation of margins versus growth. As Cherny noted, "Capacity is a resource we manage thoughtfully." In the 2026 AI landscape, the era of subsidized, unlimited compute for third-party automation is over. For the average user on Claude.ai, the experience remains unchanged; for the power users running autonomous offices, the bell has tolled.

Read more of this story at Slashdot.

https://slashdot.org/story/26/04/04/1919236/anthropic-announces-claude-subscribers-must-now-pay-extra-to-use-openclaw?utm_source=rss1.0mainlinkanon&utm_medium=feed

Posted by EditorDavid

https://it.slashdot.org/story/26/04/04/0638246/amazon-must-negotiate-with-first-warehouse-workers-union-us-labor-board-rules?utm_source=rss1.0mainlinkanon&utm_medium=feed

Amazon "must negotiate with a labor union representing some 5,000 workers at a company warehouse on Staten Island," reports Reuters, citing a ruling Wednesday from America's National Labor Relations Board (NLRB). The union formed in 2022, according to the article, and "has been seeking to negotiate with Amazon over pay, working conditions and other matters." The NLRB said in its ruling that Amazon "has engaged in unfair labor practices" by refusing to bargain with the labor group or to recognize its legitimacy... Amazon said on Thursday it disagreed with the NLRB's ruling. "Representatives of the NLRB improperly influenced this election," the company said in a statement, suggesting it planned to appeal. "We're confident an unbiased court will overturn the original certification, and we look forward to the opportunity for our team to fairly voice their opinions." An appeal would likely preclude Amazon from having to comply with the NLRB's order while it makes its way through the courts... Related to the Staten Island case, Amazon has argued that the NLRB itself is unconstitutional and sued to block the agency from ruling on it. The matter is still pending. After forming independently, that union "has since aligned with the International Brotherhood of Teamsters," the article points out. The Teamsters represent 1.3 million American workers, according to a statement they issued this week, which also includes this quote from the president of Amazon Labor Union-e Local 1. "We are making history at Amazon, and we are doing it through undiluted worker power..." Their statement adds that the ruling "came only one day after the union announced another historic victory that upheld Amazon Teamsters' right to strike."

Read more of this story at Slashdot.

https://it.slashdot.org/story/26/04/04/0638246/amazon-must-negotiate-with-first-warehouse-workers-union-us-labor-board-rules?utm_source=rss1.0mainlinkanon&utm_medium=feed

Posted by EditorDavid

https://news.slashdot.org/story/26/04/04/0334217/the-document-foundation-removes-dozens-of-collabora-developers?utm_source=rss1.0mainlinkanon&utm_medium=feed

Long-time GNOME/OpenOffice.org/LibreOffice contributor Michael Meeks is now general manager of Collabora Productivity. And earlier this month he complained when LibreOffice decided to bring back its LibreOffice Online project, as reported by Neowin, which had been inactive since 2022. After the original project went dormant — to which Collabora was a major contributor — they forked the code and created their own product, Collabora Online. But this week Meeks blogged about even more changes, writing that the Document Foundation (the nonprofit behind LibreOffice) "has decided to eject from membership all Collabora staff and partners. That includes over thirty people who have contributed faithfully to LibreOffice for many years." Meeks argues the ejections were "based on unproven legal concerns and guilt by association." This includes seven of the top ten core committers of all time (excluding release engineers) currently working for Collabora Productivity. The move is the culmination of TDF losing a large number of founders from membership over the last few years with: Thorsten Behrens, Jan 'Kendy' Holesovsky, Rene Engelhard, Caolan McNamara, Michael Meeks, Cor Nouws and Italo Vignoli no longer members. Of the remaining active founders, three of the last four are paid TDF staff (of whom none are programming on the core code). The blog It's FOSS calls it "LibreOffice Drama." They've confirmed the removals happened, also noting recently adopted Community Bylaws requiring members to step down if they're affiliated with a company in an active legal dispute with the Foundation. But The Documentation Foundation "also makes clear that a membership revocation is not a ban from contributing, with the project remaining open to anyone, and expects Collabora to keep contributing 'when the time comes.'" Collabora's Meeks adds in his blog post that there's "bold and ongoing plans to create an entirely new, cut-down, differentiated Collabora Office for users that is smoother, more user friendly, and less feature dense than our Classic product (which will continue to be supported for years for our partners). This gives a chance to innovate faster in a separate place on a smaller, more focused code-base with fewer build configurations, much less legacy, no Java, no database, web-based toolkit and more. We are excited to get executing on that. To make this process easier, and to put to bed complaints about having our distro branches in TDF gerrit [for code review], and to move to self-hosted FOSS tooling we are launching our own gerrit to host our existing branch of core... We will continue to make contributions to LibreOffice where that makes sense (if we are welcome to), but it clearly no longer makes much sense to continue investing heavily in building what remains of TDF's community and product for them — while being excluded from its governance. In this regard, we seem to be back where we were fifteen years ago.

Read more of this story at Slashdot.

https://news.slashdot.org/story/26/04/04/0334217/the-document-foundation-removes-dozens-of-collabora-developers?utm_source=rss1.0mainlinkanon&utm_medium=feed

Posted by BeauHD

https://science.slashdot.org/story/26/04/03/2334204/cognitive-surrender-leads-ai-users-to-abandon-logical-thinking-research-finds?utm_source=rss1.0mainlinkanon&utm_medium=feed

An anonymous reader quotes a report from Ars Technica: When it comes to large language model-powered tools, there are generally two broad categories of users. On one side are those who treat AI as a powerful but sometimes faulty service that needs careful human oversight and review to detect reasoning or factual flaws in responses. On the other side are those who routinely outsource their critical thinking to what they see as an all-knowing machine. Recent research goes a long way to forming a new psychological framework for that second group, which regularly engages in "cognitive surrender" to AI's seemingly authoritative answers. That research also provides some experimental examination of when and why people are willing to outsource their critical thinking to AI, and how factors like time pressure and external incentives can affect that decision. Overall, across 1,372 participants and over 9,500 individual trials, the researchers found subjects were willing to accept faulty AI reasoning a whopping 73.2 percent of the time, while only overruling it 19.7 percent of the time. The researchers say this "demonstrate[s] that people readily incorporate AI-generated outputs into their decision-making processes, often with minimal friction or skepticism." In general, "fluent, confident outputs [are treated] as epistemically authoritative, lowering the threshold for scrutiny and attenuating the meta-cognitive signals that would ordinarily route a response to deliberation," they write. These kinds of effects weren't uniform across all test subjects, though. Those who scored highly on separate measures of so-called fluid IQ were less likely to rely on the AI for help and were more likely to overrule a faulty AI when it was consulted. Those predisposed to see AI as authoritative in a survey, on the other hand, were much more likely to be led astray by faulty AI-provided answers. Despite the results, though, the researchers point out that "cognitive surrender is not inherently irrational." While relying on an LLM that's wrong half the time (as in these experiments) has obvious downsides, a "statistically superior system" could plausibly give better-than-human results in domains such as "probabilistic settings, risk assessment, or extensive data," the researchers suggest. "As reliance increases, performance tracks AI quality," the researchers write, "rising when accurate and falling when faulty, illustrating the promises of superintelligence and exposing a structural vulnerability of cognitive surrender." In other words, letting an AI do your reasoning means your reasoning is only ever going to be as good as that AI system. As always, let the prompter beware.

Read more of this story at Slashdot.

https://science.slashdot.org/story/26/04/03/2334204/cognitive-surrender-leads-ai-users-to-abandon-logical-thinking-research-finds?utm_source=rss1.0mainlinkanon&utm_medium=feed