New policy brief: Companies are increasingly allowing users to customize powerful models via fine-tuning, but how does this affect built-in safety mechanisms? A collaborative research effort examines the safety risks inherent with fine-tuning of LLMs: https://
hai.stanford.edu/policy-brief-s
afety-risks-customizing-foundation-models-fine-tuning
…
ETHICS
-
Safety Risks in LLM Fine-Tuning: Policy Analysis
By
–
-
Policy Brief on AI by Leading Researchers
By
–
This policy brief was authored by @PeterHndrsn
, @lxuechen
, @jurafsky
, @tatsu_hashimoto
, Mark A. Lemley, and @percyliang
: -
Blockchain emerges as solution to AI trust and bias problems
By
–
The difficulty of ensuring #artificialintelligence algorithms are safe, unbiased and accurate is turning some companies to a technology that has so-far failed to catch on in business: #blockchain. https://
wsj.com/articles/ai-ha
s-a-trust-problem-can-blockchain-help-ba3b26f7?st=yxlzhbdj07kfiqk
… @IsabelleBiscuit -
OpenAI implements restrictions on gambling-related requests
By
–
Oui ! 🙂 Ca a commencé coté OpenAI de bloquer les demandes de jeux d'argents.
-
AI Companions: Emotional Connection Economics and Intimacy Evolution
By
–
#Sextech aside, generative AI expands artificial intimacy into the realm of emotional connection: AI companions will be friends, lovers, coaches, and mentors. @Rahafharfoush introduces you to the players—and economics—behind this shift : atelier.net/insights/artific…
-
Startup Pitfalls: Solving Too Much, Overusing AI Branding
By
–
Startup Pitfalls
Trying to solve too much at once, overuse of "AI" branding without substance, misalignment among founders. -
Privacy and data synchronization concerns in AI model training
By
–
Even if you turn it off along with the history, it also doesn’t sync across devices as a setting. Meaning, opening your account on another device can dump your data for training before you turn it off manually
-
OpenAI Anthropic Cohere Meet Chinese AI Experts on Policy
By
–
SCOOP Employees of OpenAI, Anthropic, Cohere met with Chinese AI experts in Geneva to find common ground on AI policy and safety
-
Stanford scholars propose two-pronged AI copyright approach
By
–
As legal battles on AI copyright issues make headlines this week, here’s a refresh of last year’s November brief, where @Stanford scholars offered an analysis of the U.S. case law and recommended a two-pronged approach in addressing copyright issues. https://
stanford.io/48twY2G -
Preventing New Product Development: Policy Approaches
By
–
how do you propose we prevent people from building new products?