Anthropic Launches Claude Opus 4.5, Claiming Top Internal Safety Scores

·7 views

Anthropic has released a new artificial intelligence model it calls its strongest yet on internal safety evaluations. The model is designed to be both faster and more cost-effective than its predecessors. Anthropic stated that the new release outperformed previous models on its most rigorous in-house safety benchmarks. The launch comes amid ongoing industry-wide debate over AI safety standards and practices.

Read the full story at NYT Technology

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

US calls Australia's algorithm opt-out bill censorship, warns of tech fines

Australia has proposed new laws that would require technology companies to give users the option to disable algorithmic content recommendations. Firms that fail to comply would face financial penalties under the draft legislation. The United States government has criticised the proposal, labelling it a form of censorship. The move reflects growing international tension over how governments regulate the role of algorithms in shaping online content. Australia's push is part of a broader effort by regulators worldwide to increase platform accountability and user control.

0 comments Read more at BBC Tech

TechnologyThe Verge ·

OpenAI Forms Mathematician Panel After Backlash Over AI Math Announcements

OpenAI has announced the formation of an independent panel of mathematicians to advise the company on its engagement with the mathematics research community. The panel, revealed on Monday, will guide how AI companies present and release new mathematical results. The move follows a reputational crisis stemming from OpenAI's handling of recent high-profile math breakthroughs. While researchers have welcomed the initiative as a positive first step, many say they were caught off guard by the announcement and still have unanswered questions about the panel's scope and authority.

0 comments Read more at The Verge

TechnologyTechCrunch ·

TechCrunch Founder Summit to Cover Fundraising, Hiring and AI in Boston Nov. 4

TechCrunch is hosting its Founder Summit in Boston on November 4, targeting early-stage startup founders. The event is designed to help entrepreneurs navigate common challenges in building a company. Key topics on the agenda include fundraising strategies, hiring practices, and artificial intelligence insights. The summit aims to provide founders with practical knowledge to avoid costly mistakes and accelerate growth.

0 comments Read more at TechCrunch

TechnologyBBC Tech ·

Data Centres Turn to Light-Based Tech to Cut Energy Use and Boost Capacity

Data centres are exploring a shift from traditional copper wiring to light-based technology in a bid to reduce power consumption. The move comes as energy demand at data centres continues to grow, putting pressure on operators to find more efficient solutions. Light-speed data transmission is seen as a promising alternative that could both lower electricity use and increase network capacity. The transition reflects a broader industry push toward sustainable and scalable infrastructure for handling rising data loads.

0 comments Read more at BBC Tech