News

Friday Squid Blogging: The Reproductive Habits of Giant Squid

Interesting:

A recent study on giant squid that have washed ashore along the Sea of Japan coast has raised the possibility that the animal has a different reproductive method than many other types of squid.

Almost all squid and octopus species are polygamous, with multiple males passing sperm to a single female. Giant squids were thought to have a similar form reproduction.

However, a group led by Professor Noritaka Hirohashi, 57, a professor of reproductive biology in the Faculty of Life and Environmental Sciences at Shimane University suspects differently…

#StopRansomware: Daixin Team

Original release date: October 21, 2022CISA, the Federal Bureau of Investigation (FBI), and the Department of Health and Human Services (HHS) have released a joint Cybersecurity Advisory (CSA), #StopRansomware: Daixin Team to provide information on the…

Adversarial ML Attack that Secretly Gives a Language Model a Point of View

Machine learning security is extraordinarily difficult because the attacks are so varied—and it seems that each new one is weirder than the next. Here’s the latest: a training-time attack that forces the model to exhibit a point of view: Spinning Language Models: Risks of Propaganda-As-A-Service and Countermeasures.”

Abstract: We investigate a new threat to neural sequence-to-sequence (seq2seq) models: training-time attacks that cause models to “spin” their outputs so as to support an adversary-chosen sentiment or point of view—but only when the input contains adversary-chosen trigger words. For example, a spinned summarization model outputs positive summaries of any text that mentions the name of some individual or organization…

Adversarial ML Attack that Secretly Gives a Language Model a Point of View

Machine learning security is extraordinarily difficult because the attacks are so varied—and it seems that each new one is weirder than the next. Here’s the latest: a training-time attack that forces the model to exhibit a point of view: Spinning Language Models: Risks of Propaganda-As-A-Service and Countermeasures.”

Abstract: We investigate a new threat to neural sequence-to-sequence (seq2seq) models: training-time attacks that cause models to “spin” their outputs so as to support an adversary-chosen sentiment or point of view—but only when the input contains adversary-chosen trigger words. For example, a spinned summarization model outputs positive summaries of any text that mentions the name of some individual or organization…

Battle with Bots Prompts Mass Purge of Amazon, Apple Employee Accounts on LinkedIn

On October 10, 2022, there were 576,562 LinkedIn accounts that listed their current employer as Apple Inc. The next day, half of those profiles no longer existed. A similarly dramatic drop in the number of LinkedIn profiles claiming employment at Amazon comes as LinkedIn is struggling to combat a significant uptick in the creation of fake employee accounts that pair AI-generated profile photos with text lifted from legitimate users.