Microsoft details Skeleton Key, a new jailbreak technique in which a threat actor can convince an AI model to ignore its built-in safeguards and respond to requests for harmful, illegal, or offensive requests that might otherwise have been refused.
The post Skeleton Key the Latest Jailbreak Threat to AI Models: Microsoft appeared first on Security Boulevard.
Source: www.databreachtoday.com – Author: 1 Artificial Intelligence & Machine Learning , Next-Generation Technologies & Secure Development Microsoft Dubs the Technique ‘Skeleton Key’ Akshaya Asokan (asokan_akshaya) • June 27, 2024 In a “Skeleton Key” attack, researchers say the magic words necessary to make chatbots ignore safety guidelines. (Image: Shutterstock) Artificial intelligence researchers say they […]
La entrada Chatbots Will Break Guardrails If the Info Is ‘Educational’ – Source: www.databreachtoday.com se publicó primero en CISO2CISO.COM & CYBER SECURITY GROUP.
Microsoft has tricked several gen-AI models into providing forbidden information using a jailbreak technique named Skeleton Key.
The post Microsoft Details ‘Skeleton Key’ AI Jailbreak Technique appeared first on SecurityWeek.
Enlarge (credit: Diamond Shruumz)
Cases of illnesses linked to microdosing candies have more than doubled, with reports of seizures and the need for intubation, mechanical ventilation, and intensive care stays. But, there remains no recall of the products—microdosing chocolates, gummies, and candy cones by Diamond Shruumz—linked to the severe and life-threatening illnesses. In the latest update from the Food and Drug Administration late Tuesday, the agency said that it "has been in contact with the firm about a possible voluntary recall, but these discussions are still ongoing."
In the update, the FDA reported 26 cases across 16 states, up from 12 cases in eight states last week. Of the 26 reported cases, 25 sought medical care and 16 were hospitalized. No deaths have been reported.
Last week, the Centers for Disease Control and Prevention released a health alert about the candies. The agency noted that as of June 11, the people sickened after eating Diamond Shruumz candies presented to health care providers with a host of severe symptoms. Those include: central nervous system depression with sedation, seizures, muscle rigidity, clonus (abnormal reflex responses), tremor, abnormal heart rate (bradycardia or tachycardia), abnormal blood pressure (hypotension or hypertension), gastrointestinal effects (nausea, vomiting, or abdominal pain), skin flushing, diaphoresis (excessive sweating), and metabolic acidosis with increased anion gap (an acid-based disorder linked to poisonings).
Enlarge / Diamond Shruumz's "extremely potent" infused cones in "sprinkles" flavor. (credit: Diamond Shruumz)
More people have reported severe poisonings in an ongoing outbreak marked by people seizing and needing to be intubated after consuming microdose candies made by Diamond Shruumz, the Food and Drug Administration reported Tuesday.
There are now at least 12 reported cases across eight states. All 12 people were ill enough to seek medical care, and 10 needed to be hospitalized. The symptoms reported so far include seizures, central nervous system depression (loss of consciousness, confusion, sleepiness), agitation, abnormal heart rates, hyper/hypotension, nausea, and vomiting, the FDA reported.
In Tuesday's update, the FDA also expanded the products linked to the illnesses. In addition to all flavors of Diamond Shruumz's Microdosing Chocolate Bars, the agency's warning now covers all flavors of the brand's Infused Cones and Micro Dose and Macro Dose Gummies.
Enlarge / A Diamond Shruumz chocolate bar, which come in a variety of flavors. (credit: diamondshruumz.com)
Various federal and state health officials are sounding the alarm on Diamond Shruumz-brand Microdosing Chocolate Bars. The candy, said to be infused with mushrooms, has been linked to severe illnesses, including seizures, loss of consciousness, confusion, sleepiness, agitation, abnormal heart rates, hyper/hypotension, nausea, and vomiting, according to an outbreak alert released by the Food and Drug Administration on Friday.
So far, eight people across four states have been sickened—four in Arizona, two in Indiana, one in Nevada, and one in Pennsylvania, the FDA reported. Of the eight, six have been hospitalized.
"We are urging the public to use extreme caution due to the very serious effects of these products," Maureen Roland, director of the Banner Poison and Drug Information Center in Phoenix, said in a press release earlier this week.
Enlarge (credit: gmast3r | iStock / Getty Images Plus)
After the most downloaded local news app in the US, NewsBreak, shared an AI-generated story about a fake New Jersey shooting last Christmas Eve, New Jersey police had to post a statement online to reassure troubled citizens that the story was "entirely false," Reuters reported.
"Nothing even similar to this story occurred on or around Christmas, or even in recent memory for the area they described," the cops' Facebook post said. "It seems this 'news' outlet's AI writes fiction they have no problem publishing to readers."
It took NewsBreak—which attracts over 50 million monthly users—four days to remove the fake shooting story, and it apparently wasn't an isolated incident. According to Reuters, NewsBreak's AI tool, which scrapes the web and helps rewrite local news stories, has been used to publish at least 40 misleading or erroneous stories since 2021.
Login
