Breaking AI & Tech News Analyzed

The Guardian released a new podcast episode titled The AI jailbreakers, where journalist Jamie Bartlett sits down with researcher Annie Kelly to dissect the underground movement that tests the boundaries of today’s most advanced chatbots.Podcast Uncovers the Tactics Behind AI JailbreaksIn the hour‑long conversation, Bartlett and Kelly map out how actors exploit prompts, system messages, and external tools to coax models such as ChatGPT, Gemini, Grok and Claude into producing prohibited content. They highlight three core techniques:Prompt engineering: chaining innocuous queries to bypass safety filters.Context injection: feeding the model with fabricated system instructions that override its guardrails.Tool‑assisted loops: using APIs or browser extensions to automate repeated jailbreak attempts.Scale of Jailbreak Attempts and Model VulnerabilitiesWhile exact numbers are scarce, the hosts cite recent research indicating:Over 10,000 distinct jailbreak prompts have been catalogued across major LLMs in the past year.Success rates vary by model, with open‑source variants showing 30‑40% higher breach rates than proprietary systems.Each successful breach can expose hundreds of megabytes of filtered training data or generate disallowed content at scale.Why Jailbreaks Threaten Trust in Generative AIThe discussion moves beyond technical tricks to the broader societal stakes. Unchecked jailbreaks can:Facilitate the spread of hate speech, extremist propaganda, or illegal instructions.Erode user confidence, prompting regulators to impose stricter compliance regimes.Accelerate an arms race between jailbreakers and AI developers, diverting resources from innovation to defense.Future of AI Safety: Anticipating the Next Wave of Jailbreak DefensesBoth guests agree that the next phase will involve layered defenses:Dynamic safety layers: real‑time monitoring that adapts to emerging jailbreak patterns.Transparency dashboards: public logs of attempted breaches to inform policy and research.Collaborative bounty programs: incentivizing ethical hackers to report vulnerabilities before malicious actors exploit them.As AI systems become more embedded in daily life, understanding the mindset of jailbreakers will be crucial for building resilient, trustworthy models.

The Justice Department has launched a historic legal assault on the Southern Poverty Law Center (SPLC), accusing the renowned civil rights organization of a $3 million fraud scheme involving informants within extremist groups. Acting Attorney General Todd Blanche alleges that the SPLC defrauded donors by using their contributions to fund the very extremism it claimed to be dismantling, a stark contradiction of its mission.Key DevelopmentsAlleged Scheme: The DOJ claims the SPLC raised millions through a secret informant program, paying individuals affiliated with groups like the Ku Klux Klan and the National Socialist Party of America.Timeline: Payments occurred between 2014 and 2023, totaling at least $3 million.Charges: The civil rights group faces federal charges including wire fraud, bank fraud, and conspiracy to commit money laundering.Operational Details: Prosecutors allege funds were routed through two bank accounts and loaded onto prepaid cards before being distributed to informants, some of whom were high-ranking members of neo-Nazi organizations.Data & Market ImpactThe $3 million figure represents a significant portion of the SPLC's operational budget, raising questions about the financial transparency of large non-profit watchdog organizations. This lawsuit marks a rare instance of the federal government targeting a major civil rights organization, potentially setting a precedent for how future administrations handle non-profit accountability. The legal action comes at a time when the non-profit sector is already navigating increased scrutiny regarding political bias and financial mismanagement.Why This MattersThis case strikes at the heart of civil rights monitoring in the United States. If the allegations are proven true, they would severely undermine the credibility of an organization that has served as a primary resource for law enforcement and the public regarding hate groups. For donors and the public, the revelation that funds intended for advocacy were used to pay informants within hate groups creates a crisis of trust. Furthermore, the timing of the lawsuit—under a Trump administration that views the SPLC as a partisan adversary—raises concerns about the weaponization of federal law enforcement against political opponents.Expert InsightThe core irony of the indictment lies in the SPLC's own admission: they were using the very tactics of infiltration and surveillance that they often criticize in others. While the group argues the program was necessary to protect lives and share intelligence with law enforcement, the DOJ frames it as a cynical cash grab. Strategically, this move by the Trump administration appears to be a two-pronged attack: it attempts to discredit a powerful liberal watchdog organization while simultaneously signaling a hardline stance against extremism. By targeting the SPLC, the administration may be attempting to delegitimize the broader discourse on hate speech and domestic terrorism.What Happens NextThe SPLC has vowed a vigorous defense, with CEO Bryan Fair stating the organization will fight to protect its staff and mission. The coming months will likely see intense legal battles that could set a major precedent for non-profit transparency. If the DOJ prevails, it could lead to stricter regulations on how civil rights organizations handle informant programs and donor funds. Conversely, a successful defense by the SPLC could reinforce the importance of undercover operations in combating hate groups, though it would likely do little to quell the political polarization surrounding the case.

In a vivid display of dissent, demonstrators assembled in Paris to condemn a draft law described by opponents as a ‘liberty‑killing’ anti‑Semitism bill. The rally underscored growing concerns that the legislation could erode civil liberties while attempting to combat hate speech. Participants voiced their alarm that the bill’s provisions might set a precedent for broader restrictions on free expression, urging lawmakers to reconsider its wording.

TikTok announced the removal of a high‑profile Israeli far‑right influencer’s account after The Guardian identified videos in which he assaulted left‑wing activists in the occupied West Bank. The platform said the content breached its community guidelines on hate speech and bullying. The influencer, identified as Roi Star, posted footage in January showing himself entering a house used by activists in Ras Ein al‑Auja and using pepper spray on a protester who tried to stop him. In the same clip, he shouted, “This is Judea, not Palestine,” and later threatened to disclose personal details of the activists and their families. When contacted, Star claimed he was “talking about peace” and argued that the area was an open Israeli public space, insisting that his use of pepper spray was the “most minimal” defensive measure. He later described the incident as “acting” and said his intentions were not “extreme.” TikTok’s statement emphasized that its policies prohibit “violent and hateful individuals, including extremist praise or glorification,” and that the account was taken down for breaching these rules. The company also said it had removed additional videos linked to other Israeli far‑right agitators, though it did not disclose further details. The incident occurs against a backdrop of rising far‑right activity on social platforms since the Gaza war began in October 2023. Israeli forces and settlers have been responsible for the deaths of over 1,000 Palestinians in the West Bank, and recent weeks have seen an escalation of attacks on homes and activists. Human‑rights groups warn that the online amplification of such content fuels real‑world violence. Yuli Novak, executive director of B’Tselem, said dehumanising Palestinians has become “mainstream in Israel,” while digital‑media scholar Prof. Anat Ben‑David highlighted the “troubling convergence between platform dynamics and on‑the‑ground violence.” Activists on the ground report a profound psychological impact, with one resident of Masafer Yatta noting that the videos heighten fear among Palestinians living under daily settler attacks. The phenomenon mirrors the rhetoric of Israel’s far‑right politicians. In August 2025, National Security Minister Itamar Ben‑Gvir faced criticism for posting a video taunting Palestinian leader Marwan Barghouti, while Knesset member Zvi Sukkot was filmed denying settler violence in the West Bank, framing the settlement enterprise as a biblical right. Meta’s Instagram continues to host numerous accounts linked to similar agitators, though the company has not responded to requests for comment. Experts argue that while platform policies technically forbid hate‑speech, their vague language allows harmful content to spread unchecked, underscoring the need for stronger enforcement to curb the digital propagation of extremist narratives.

Ethiopian women's rights activists are facing a rising tide of digital violence, including online threats, doxing, and deepfake abuse, forcing some to flee the country. Yordanos Bezabih, an Ethiopian women's rights activist, had faced online threats for years, including acid attacks, gang-rape, and death. However, in 2025, the threats became more menacing, with an anonymous Telegram group organizing an effort to track down her location.The group shared deepfakes of her – nude images and videos. A stranger started filming her in the streets, calling her by her social media handle. Thieves broke into her house and stole her laptop. Soon after, her Telegram account was hacked, and her private photos and messages were circulated on social media. The perpetrators later circulated her address, demanding she be found and “executed”.In August, Bezabih left Ethiopia on a fellowship for human rights defenders. She has not returned since; it is too dangerous. “I have been forced to remain outside the country in order to protect my safety and continue my work,” she says.Bezabih is one of a small but growing number of feminists and women’s rights defenders who have left Ethiopia over the past two years, as online violence has become all-pervasive and uncontrolled. Three years after Facebook was accused of allowing hate speech to spread unchecked in Ethiopia, amid genocidal violence against ethnic Tigrayans during the civil war – claims rejected by Meta – social media inciters in Ethiopia have found a new target: women online.Research by the Centre for Information Resilience (CIR) bears out the scale of online gendered abuse in Ethiopia. Its 2024 report, Silence, Shamed and Threatened, found that technology-facilitated gender-based violence (TFGBV) has become “normalized to the point of invisibility” and is a daily occurrence with severe offline impacts, including psychological harm, physical assault, and arrests.Activists say the government and social media platforms are not doing enough to protect them. “I don’t think the government is much concerned about online harassment. It is barely a government agenda,” says Befekadu Hailu, an Ethiopian civil society leader and former director of Ethiopia’s Centre for the Advancement of Rights and Democracy.Bezabih says the online platforms that enable the violence also do little about it. “Even though they claim to have all these community guidelines, tech platforms never respond to reports, claims or even appeals.”

Australia's recent introduction of 'hate speech' laws has sparked controversy, with human rights groups warning that the legislation could be used to suppress legitimate criticism of Israel's actions in Palestine. The laws, which were rushed through the New South Wales federal parliament in January, specifically target speech considered anti-Semitic and carry a maximum penalty of two years' imprisonment for those found guilty. Arif Hussein, senior lawyer at the Human Rights Law Centre, has expressed concerns that the legislation could be used to unfairly impact peaceful protest and speech regarding Israel's actions in Palestine. “There are serious concerns that the new laws could have a chilling effect on legitimate activism and protest and unfairly impact peaceful protest and speech regarding Israel’s actions in Palestine,” Hussein said. The laws have been introduced in response to a mass shooting at Sydney's Bondi Beach in December, which killed 15 people. However, critics argue that the legislation prioritizes the protection of the Jewish community over other vulnerable groups. Greens party spokesperson Senator David Shoebridge has criticized the laws, saying they were 'deliberately designed to not protect' vulnerable communities such as Muslims, women, and LGBTQ+ Australians. “Unfortunately, the Albanese Labor government and the Liberals made the decision to protect just one religion,” he said. The laws have also been criticized for their broad and poorly defined criminal powers, which could be used against legitimate human rights groups, including those focused on Palestine. Police powers across Australia have also been strengthened to crack down on pro-Palestine protests, with reports of excessive force used against demonstrators. The controversy surrounding the laws highlights the challenges of balancing free speech with the need to combat hate speech and protect vulnerable communities.

A 60-year-old man has received a suspended prison sentence after pleading guilty to sending offensive messages about England international Jess Carter during the European Championship last summer.Nigel Dewale, of Great Harwood in Lancashire, was sentenced to six weeks, suspended for 12 months, after a hearing at Blackburn magistrates court. He also received a four-year football banning order and a requirement to engage in a 10-day rehabilitation programme.Dewale had admitted to sending 'malicious communications' on TikTok and was identified as the author of two posts that used both racist and misogynist language.The deputy director of the UK Football Policing Unit, Mike Ankers, welcomed the verdict and praised the determination of Carter and the rest of the Lionesses to stand up to online abuse.Ankers said he believed there continued to be substantial underreporting of the online abuse directed at athletes and called for social media companies to be held accountable for sharing identifying information of account holders suspected of hate speech.