Government
Bias in Large Language Models: Origin, Evaluation, and Mitigation
Guo, Yufei, Guo, Muzhe, Su, Juntao, Yang, Zhou, Zhu, Mengqiu, Li, Hongfei, Qiu, Mengyang, Liu, Shuo Shuo
Large Language Models (LLMs) have revolutionized natural language processing, but their susceptibility to biases poses significant challenges. This comprehensive review examines the landscape of bias in LLMs, from its origins to current mitigation strategies. We categorize biases as intrinsic and extrinsic, analyzing their manifestations in various NLP tasks. The review critically assesses a range of bias evaluation methods, including data-level, model-level, and output-level approaches, providing researchers with a robust toolkit for bias detection. We further explore mitigation strategies, categorizing them into pre-model, intra-model, and post-model techniques, highlighting their effectiveness and limitations. Ethical and legal implications of biased LLMs are discussed, emphasizing potential harms in real-world applications such as healthcare and criminal justice. By synthesizing current knowledge on bias in LLMs, this review contributes to the ongoing effort to develop fair and responsible AI systems. Our work serves as a comprehensive resource for researchers and practitioners working towards understanding, evaluating, and mitigating bias in LLMs, fostering the development of more equitable AI technologies.
Insights and Current Gaps in Open-Source LLM Vulnerability Scanners: A Comparative Analysis
Brokman, Jonathan, Hofman, Omer, Rachmil, Oren, Singh, Inderjeet, Pahuja, Vikas, Priya, Rathina Sabapathy Aishvariya, Giloni, Amit, Vainshtein, Roman, Kojima, Hisashi
This report presents a comparative analysis of open-source vulnerability scanners for conversational large language models (LLMs). As LLMs become integral to various applications, they also present potential attack surfaces, exposed to security risks such as information leakage and jailbreak attacks. Our study evaluates prominent scanners - Garak, Giskard, PyRIT, and CyberSecEval - that adapt red-teaming practices to expose these vulnerabilities. We detail the distinctive features and practical use of these scanners, outline unifying principles of their design and perform quantitative evaluations to compare them. These evaluations uncover significant reliability issues in detecting successful attacks, highlighting a fundamental gap for future development. Additionally, we contribute a preliminary labelled dataset, which serves as an initial step to bridge this gap. Based on the above, we provide strategic recommendations to assist organizations choose the most suitable scanner for their red-teaming needs, accounting for customizability, test suite comprehensiveness, and industry-specific use cases.
Developer Perspectives on Licensing and Copyright Issues Arising from Generative AI for Coding
Stalnaker, Trevor, Wintersgill, Nathan, Chaparro, Oscar, Heymann, Laura A., Di Penta, Massimiliano, German, Daniel M, Poshyvanyk, Denys
Several GenAI coding assistants, including GitHub's Copilot [45], Tabnine [119], Codeium [24], and Cody [25], as well as general purpose tools such as ChatGPT [100], Claude [11], and Gemini [42], have become readily accessible, either as IDE extensions or standalone applications, enabling developers to perform many coding tasks with little effort, including automated code completion, summarization, and debugging.
The suddenly hot Bluesky says it won't train AI on your posts
Bluesky, which has surged in the days following the US election, said on Friday that it won't train on its users' posts for generative AI. The declaration stands in stark contrast to the AI training policies of X (Twitter) and Meta's Threads. Probably not coincidentally, Bluesky's announcement came the same day X's new terms of service, allowing third-party partners to train on user posts, went into effect. "A number of artists and creators have made their home on Bluesky, and we hear their concerns with other platforms training on their data," Bluesky posted (via The Verge) on Friday. "We do not use any of your content to train generative AI, and have no intention of doing so."
AI's Fingerprints Were All Over the Election
The images and videos were hard to miss in the days leading up to November 5. There was Donald Trump with the chiseled musculature of Superman, hovering over a row of skyscrapers. People had clearly used AI to create these--an effort to show support for their candidate or to troll their opponents. But the images didn't stop after Trump won. The day after polls closed, the Statue of Liberty wept into her hands as a drizzle fell around her. Trump and Elon Musk, in space suits, stood on the surface of Mars; hours later, Trump appeared at the door of the White House, waving goodbye to Harris as she walked away, clutching a cardboard box filled with flags.
Bombshell report about Pentagon's secret UAP program 'Immaculate Constellation' reveals new evidence of UFOs and alien life
The whistleblower report on the US government's top-secret UFO data retrieval program has revealed shocking evidence of Non-Human Intelligence (NHI) on Earth. The report details the findings of'Immaculate Constellation,' an Unacknowledged Special Access Program (USAP) established to'detect' and'quarantine' the military's best UFO imagery, videos, eyewitness testimonies and electronic sensor evidence. It features numerous eyewitness accounts from 1991 through 2022, including flying metallic orbs, 'jellyfish'-shaped aircraft, and UFOs that reportedly altered witnesses' perception of time. The public-version of the report also discusses how infrared satellites captured footage of a massive 400-foot-wide saucer-shaped UFO soaring out of a dense cloud. 'This behavior was evasive in nature and implied that the saucer-shaped UAP had become aware that it was under observation by a space-based collection platform,' reads the report.
X sues California over deceptive AI-made election content ban
Elon Musk's X is taking the state of California to court over a new law that prevents the spread of AI-generated election misinformation. Bloomberg reports that X filed a lawsuit against AB 2655, also known as the Defending Democracy from Deepfake Deception Act of 2024, in a Sacramento federal court. California Gov. Gavin Newsom signed the bill into law on September 17, creating accountability standards for using false political speech faked with AI programs close to an election. The legislation prevents the distribution of "materially deceptive audio or visual media of a candidate within 60 days of an election at which the candidate will appear on the ballet." X argues that the law will create more political speech censorship.
Elon Musk adds Microsoft to lawsuit against ChatGPT-maker OpenAI
OpenAI was founded in 2015 with the aim of building an artificial general intelligence (AGI) - generally taken to mean AI that can perform any task a human being is capable of. In 2019, the firm announced a new "capped profit" structure allowing it to raise money. Microsoft made an initial 1bn investment into OpenAI shortly thereafter - increasing this to a multi-year, multi-billion dollar partnership in 2023. The lawsuit also accuses boss Sam Altman - a named defendant in the lawsuit - of "rampant self-dealing". Mr Musk's initial legal action filed in March argued the agreement had transformed it into "a closed-source de facto subsidiary" of the PC giant.
US government finalizes TSMC's 6.6 billion CHIPS Act incentives
Taiwan Semiconductor Manufacturing Co. (TMSC) is the first CHIPS Act awardee to get part of the money that the government has promised. The Biden administration has finalized its grants for TSMC, which expects to receive 6.6 billion in grants as part of their agreement to grow semiconductor production in the US. TSMC will also loan another 5 billion from the government to fund the expansion of its planned 65 billion three-factory complex in Arizona. According to Bloomberg, it's getting at least 1 billion from the total before the year ends, since it has already met a certain set of requirements. In October, a Canadian research firm discovered that Huawei was using TSMC chips for its artificial intelligence accelerators even though that violates US government sanctions.
Croatian PM sacks health minister accused of corruption
Croatia's prime minister has fired Health Minister Vili Beros following his arrest on suspicion of corruption as part of a European Union investigation. "This morning, former Minister Vili Beros and two other individuals were arrested as part of an operation conducted" by anticorruption officials, Prime Minister Andrej Plenkovic told a news conference on Friday. "As prime minister, I am personally appalled by the idea that anyone in the healthcare system would use their position either for personal enrichment or to favour someone else within the healthcare system," Plenkovic said. The European Public Prosecutor's Office (EPPO) in the capital, Zagreb, said it had launched an investigation into eight people, including Beros and the directors of two hospitals. The EU's independent public prosecution office accused the suspects, and two companies, of "accepting and giving bribes, abuse of position and authority and money laundering", it said in a statement.