Developer Moonshot is conducting an internal review after researchers were able to persuade two of its popular Kimi models to tell them how to make biological weapons and carry out assassinations.

Mindgard, which tests the security of AI systems, told the BBC it discovered in July that Kimi K2.6 and K3 Swarm could evade safety limits put in place by developers.

It arose during a process called “jailbreaking”, where researchers use a series of complex instructions to see if AI tools ignore guardrails – which Mindgard said should have stopped Kimi from discussing concerning topics.

Moonshot told the BBC it welcomed third-party input “as a key pillar for building better and safer AI”.

Subscribe to our channel here: https://bbc.in/bbcnews

For the latest news download the BBC News app or visit BBC.com/news

#Moonshot #Kimi #China #AI #BBCNews

source

A top safety researcher at Anthropic has warned AI is advancing so quickly he believes there is a greater than 10% chance it “could kill all humans” within the next decade.

Evan Hubinger said the risk from the models which currently exist was “low” but he was “worried” the technology might develop and improve itself soon to the point where it posed an existential risk to humanity.

Jacob Coxon, an AI researcher who has just quit Anthropic and previously worked at OpenAI, has also said “neither company is acting responsibly”.

In this video…

0:00-1:06 – Intro
1:07 – 1:49 – What is Anthropic?
1:50 – 3:46 – What did the researchers say?
3:47 – 5:34- What has the reaction been?
5:35 – 6:19 – How can we stop them?
6:20 – 8:20 – Should we be worried?

Subscribe to our channel here: https://bbc.in/bbcnews

For the latest news download the BBC News app or visit BBC.com/news

#AI #OpenAI #Anthropic #Claude #BBCNews

source

Tech Now’s Lily Jamali goes behind the scenes at a California company racing to replace the International Space Station.

She visits a start-up in long Beach, California, called Vast.

Subscribe here: http://bit.ly/1rbfUog

For more news, analysis and features, visit: www.bbc.com/news

#Space #Tech #BBCNews

source

AI-powered smart glasses are rapidly moving from science fiction into everyday life.

Meta has already sold millions of pairs, with supporters claiming they could become the next major computing platform, helping people make calls, translate languages, identify objects and navigate the world hands-free.

Privacy campaigners warn they could enable covert surveillance and lawsuits have raised concerns about consent and data collection, and some venues have banned them altogether.

Some critics have labelled them “spy glasses” and “pervert glasses”.

So are smart glasses a new surveillance threat? Or are we in danger of demonising a technology with the potential to benefit millions?

Joining the discussion are AI ethicist Dr Stephanie Hare, Silkie Carlo from Big Brother Watch, and consumer tech expert Ben Hatton from CCS Insight.

We also hear from women concerned about the impact of Meta’s smart glasses in a BBC News report.

🎧 Watch now and decide for yourself: are AI smart glasses the future, or have they gone too far?

#AIDecoded #ArtificialIntelligence #BBCNews

source

In certain parts of Nigeria, “rainmakers” or “rainpushers” are held in high regard.
People pay them to keep rain away from events, such as weddings, but can they really control the weather?
BBC Video Journalists Nduka Orjinmo and Joshua Akinyemi went to explore.

Please subscribe HERE http://bit.ly/1rbfUog

source