Data Poisoning
- All
- News
-
Anthropic Warns That Minimal Data Contamination Can ‘Poison’ Large AI Models
- Saturday October 11, 2025
- Written by Akash Dutta, Edited by Ketan Pratap
Anthropic has warned that even a few poisoned samples in a dataset can compromise an AI model. A joint study with the UK AI Security Institute found that as few as 250 malicious documents can implant backdoors in LLMs up to 13B parameters, proving model size offers no protection.
-
www.gadgets360.com
-
Exclusive: Tigress MT-1 Poisoned And Gone, Forest Department Says "We Don't Know"
- Tuesday July 29, 2025
- India News | Reported by Anurag Dwary
Where is Tigress MT-1 of Shivpuri's Madhav Tiger Reserve? The state forest department says it doesn't know.
-
www.ndtv.com
-
"Yamuna Water Extremely Poisonous": Arvind Kejriwal Responds To Poll Panel
- Thursday January 30, 2025
- India News | NDTV News Desk
Former Delhi Chief Minister Arvind Kejriwal, bearing the brunt of criticism over his pre-poll allegation that Haryana is poisoning waters of the Yamuna, has responded to the Election Commission with data about pollution of the river.
-
www.ndtv.com
-
Cyber Attackers Can Disable AI Systems By 'Data Poisoning': Google AI Expert
- Tuesday July 11, 2023
- Feature | Edited by Nikhil Pandey
A researcher at the conference in Shanghai claims that attackers can seriously damage artificial intelligence models by "poisoning" data sets through minor modification.
-
www.ndtv.com
-
Need Hard Data On Skripal Case To Draw Any Conclusions: Kremlin
- Friday September 28, 2018
- World News | Reuters
Moscow needs hard data from British authorities on the Skripal poisoning case and cannot draw conclusions about who was responsible based on unconfirmed media reports, Kremlin spokesman Dmitry Peskov said on Friday.
-
www.ndtv.com
-
Anthropic Warns That Minimal Data Contamination Can ‘Poison’ Large AI Models
- Saturday October 11, 2025
- Written by Akash Dutta, Edited by Ketan Pratap
Anthropic has warned that even a few poisoned samples in a dataset can compromise an AI model. A joint study with the UK AI Security Institute found that as few as 250 malicious documents can implant backdoors in LLMs up to 13B parameters, proving model size offers no protection.
-
www.gadgets360.com
-
Exclusive: Tigress MT-1 Poisoned And Gone, Forest Department Says "We Don't Know"
- Tuesday July 29, 2025
- India News | Reported by Anurag Dwary
Where is Tigress MT-1 of Shivpuri's Madhav Tiger Reserve? The state forest department says it doesn't know.
-
www.ndtv.com
-
"Yamuna Water Extremely Poisonous": Arvind Kejriwal Responds To Poll Panel
- Thursday January 30, 2025
- India News | NDTV News Desk
Former Delhi Chief Minister Arvind Kejriwal, bearing the brunt of criticism over his pre-poll allegation that Haryana is poisoning waters of the Yamuna, has responded to the Election Commission with data about pollution of the river.
-
www.ndtv.com
-
Cyber Attackers Can Disable AI Systems By 'Data Poisoning': Google AI Expert
- Tuesday July 11, 2023
- Feature | Edited by Nikhil Pandey
A researcher at the conference in Shanghai claims that attackers can seriously damage artificial intelligence models by "poisoning" data sets through minor modification.
-
www.ndtv.com
-
Need Hard Data On Skripal Case To Draw Any Conclusions: Kremlin
- Friday September 28, 2018
- World News | Reuters
Moscow needs hard data from British authorities on the Skripal poisoning case and cannot draw conclusions about who was responsible based on unconfirmed media reports, Kremlin spokesman Dmitry Peskov said on Friday.
-
www.ndtv.com