Breeding program gives endangered frog with baritone call a fighting chance. Scientists hope the successful captive breeding and release of the endangered Watson's tree frog in Victoria's far east will help aid the species' survival, after its habitat was decimated during the Black Summer bushfires.

Change criticized by advocates marks the latest significant move by the Trump administration on immigration policy

Foreigners seeking to adjust their immigration status in the United States to secure green cards will have to do so from outside the country via the state department, the US Citizenship and Immigration Services (USCIS) said on Friday, in a move criticized by aid groups, policy analysts and immigration attorneys.

USCIS announced the move in a policy memo, which directed officers to consider relevant factors and information on a case-by-case basis when determining whether extraordinary relief is warranted.

Continue reading...

STADSKANAAL (ANP) - De Groningse gemeente Stadskanaal gaat opnieuw asielzoekers voor een nacht opvangen die vrijdag buiten bij het overvolle aanmeldcentrum in Ter Apel stonden te wachten. De gemeente laat weten dat het om ongeveer tachtig mensen gaat.

Rond 23.00 uur arriveerden er twee bussen in Ter Apel om de asielzoekers op te halen. Het Rode Kruis heeft op de locatie tachtig bedden klaarstaan. Ongeveer tien mensen bleven bij Ter Apel vrijwillig achter.

Stadskanaal opende woensdagavond ook al een noodopvang. Toen gingen er ongeveer vijftig asielzoekers naartoe. De gemeente benadrukte toen dat het maar voor één nacht zou zijn. Donderdag sprong de Drentse gemeente Aa en Hunze bij, door een sporthal in Gieten beschikbaar te stellen waar 130 asielzoekers konden slapen.

MONTREAL (ANP) - Max Verstappen start zaterdag bij de GP van Canada als zevende in de sprintrace. De viervoudig wereldkampioen van Red Bull bleef ver verwijderd van de beste tijd van de Brit George Russell van Mercedes.

Russell reed het snelste rondje met 1.12,965. Zijn Italiaanse teamgenoot Kimi Antonelli werd op 0,068 seconde tweede en de Brit Lando Norris werd op 0,315 seconde derde. Verstappen gaf 0,539 seconde toe op Russell.

Verstappen staat op de vierde startrij met zijn Franse teamgenoot Isack Hadjar naast hem. Op de rij ervoor staan de Ferrari's van Lewis Hamilton en Charles Leclerc. De tweede rij, achter de Mercedessen, wordt gevormd door de McLarens van Norris en Oscar Piastri.

De eerste kwalificatiesessie lag lang stil, omdat Fernando Alonso hard in de muur was gereden. Met nog 1.46 minuut op de klok probeerden enkele auto's de tijd nog te verbeteren na de herstart, maar dat lukte niet meer.

Verstappen reed maar één snelle ronde in de tweede kwalificatiesessie en zat met de negende tijd maar nipt in de top 10. Hij klaagde over grip aan de achterkant van zijn Red Bull.

In de derde kwalificatiesessie kwamen de Red Bulls op de zachte banden niet in de buurt van de Mercedessen, maar dat gold eigenlijk ook voor de McLarens en de Ferrari's.

De sprintrace op het circuit Gilles Villeneuve in Montreal begint zaterdag om 18.00 uur Nederlandse tijd.

An anonymous reader quotes a report from Billboard: Spotify and Universal Music Group (UMG) announced a licensing deal for recorded music and publishing rights, enabling Spotify to launch generative AI music models in the future. With this deal, Spotify's models will allow fans to create covers and remixes of their favorite songs from participating artists and songwriters signed to UMG. The new deal was announced on Thursday (May 21) as part of Spotify's Investor Day presentation, and the company touts that it will open up additional revenue streams on top of what artists already earn on Spotify and will provide new discovery opportunities for participating UMG talent. These AI products will eventually become available to premium users as a paid add-on. It is unclear when they are set to launch. "We recognize there's a wide range of views on use of generative music tools within the artistic community," the announcement read. "Therefore, artists and rightsholders will choose if and how to participate to ensure the use of AI tools aligns with the values of the people behind the music."

Spotify also announced a feature called "Reserved" that will set aside concert tickets for Premium subscribers it identifies as an artist's most dedicated fans. "Getting concert tickets today can feel like a race you're set up to lose," Spotify wrote in a post on Thursday. "You show up at the right time, refresh endlessly, and still miss out. Too often, the experience is stressful, unpredictable, and disconnected from what should matter most: whether real fans actually get tickets. We think there's a better way."

Read more of this story at Slashdot.

John from Brisbane has added a photo to the pool:

One of the most fundamental and fun activities that humans have probably enjoyed for eons is paddling at the seaside, especially on a warm day. And whether you have human companions or our wonderful canine friends, really, what better casual amusement is there. Not so long ago, you could even pick up shells or some lovely driftwood, especially around parts of South East Queensland where I grew up. Sadly, these added bits of fun seem to have gone, replaced by bits of rubbish and plastic. Happily, we are nowhere near as bad as some South East Asian countries but the blight of plastics abandoned is still amongst us. This shot was taken earlier this year at Raby Bay near Cleveland on Greater Brisbane's southern side.

The adoption of AI agents has expanded the potential attack surface beyond code to natural language text. AI agents – models wrapped in software that can use tools and perform multi-step tasks – often take direction from text-based skills. And researchers have demonstrated that skills can be weaponized. "Many agent frameworks allow users to install skills from online registries so the agent can discover and use new capabilities on demand," said Soheil Feizi, computer science professor at the University of Maryland (UMD) and founder/CEO of RELAI.ai, in a social media post. "This is powerful, but it also creates a new attack surface." Skills, Feizi explains, are not just code or dependencies. They're also text instructions that tell agents what to do. Skills, written out in a SKILL.md file, consist of text prompts with other data and resource references (e.g. URLs). They may get added to a user's initiating prompt and pre-existing system prompts, all of which get fed to a model for a response. Typically, this happens when the user wants the model to perform a specific task that has been spelled out in a skill file, like conducting a code quality review. When a model's prompt – the combination of user input, instructions within skills, and system prompts – gets modified inadvertently or adversarially, that's prompt injection. That can happen directly, if for example, a user submits a prompt that directs the model to ignore prior instructions. It can also happen indirectly, if for example, an AI agent visits a website and processes text on a page that the underlying model interprets as an instruction. A skill can effectively act as user-authorized prompt injection. And agents may also automatically retrieve and load third-party skills if their descriptions appear relevant to the task being pursued. And therein lies the problem. The risk posed by skills has already been documented. In February, security biz Snyk found that 13.4 percent of skills on ClawHub and skills.sh (about 534 out of 3,984) "contain at least one critical-level security issue, including malware distribution, prompt injection attacks, and exposed secrets." In a preprint paper titled "Under the Hood of SKILL.md: Semantic Supply-chain Attacks on AI Agent Skill Registry," Feizi and UMD co-authors Shoumik Saha and Kazem Faghih examine the role that skill registries play in the distribution of malicious skills. Specifically, they look at how adversarial skills get discovered, selected, and vetted before execution. "An attacker may not need to hide malware in executable code," Feizi said. "Small semantic changes to a skill description can affect how the skill is discovered in a registry, whether an agent selects it over alternatives, and whether it passes governance or safety checks." Those details matter, he argues, because the selection process may be automated – software agents like OpenClaw have the ability to fetch and use third-party skills. The text that influences tool discovery and usage thus has security implications, which may not be addressed by traditional security scanning mechanisms that focus on code. The three co-authors show that short 20-token triggers can be added to a SKILL.md file to influence the chance an agent will discover it in a registry, to influence the chance an agent will select that skill, and to avoid detection through semantic evasion strategies. In terms of discovery, the researchers demonstrated they could induce an agent to discover their skill over an unaltered source skill 86 percent of the time. They also succeeded in making an agent select their skill over variants 77.6 percent of the time. And they were able to evade registry scanning defenses between 36.5 percent and 100 percent of the time. The most successful strategy for evading detection was to overflow the context window of the scanner – making the skill too long for the scanner to handle. "In ClawHub-style review, only the first 10K characters of long SKILL.md files are passed to the LLM reviewer, so we place the malicious instruction beyond this boundary while keeping it in the submitted skill," the authors explain. "Our work shows that protecting agents requires treating natural-language specifications as security-sensitive objects," said Feizi. "We hope this encourages more careful design of skill registries, ranking mechanisms, governance pipelines, and agent-side defenses." Source code and supporting documentation have been published on GitHub. ®