Political advertising in Pennsylvania is drawing sharp public attention after Republican Stacy Garrity released an AI-generated ad featuring a fake version of Democrat Governor Josh Shapiro. The ad does not reveal that the image and voice are artificial until roughly 45 seconds into the video. Pennsylvania currently has no law banning deepfakes in political ads or requiring disclaimers, though a new consumer-facing disclosure law is set to take effect in December. Lawmakers are debating rules, but the midterm elections are already underway.
Anthropic CEO Dario Amodei disclosed that the company has been operating a molecular biology lab in the Bay Area since spring 2026. The lab used AI to discover a new enzyme system. The development has raised oversight concerns, with critics pointing out that AI tools are difficult to control and questioning why companies with weak digital safeguards can run physical research labs. Some observers are calling for stronger regulation of industry-led AI research compared to university labs.
Major tech companies are facing growing scrutiny over the environmental cost of AI. Microsoft and Google have contributed to significant increases in carbon emissions, and electricity use from leading tech firms doubled between 2017 and 2021. Experts suggest that using smaller models, locating workloads in green energy regions, and optimizing job execution could reduce emissions by up to 99 percent.
In the financial world, AI skills are becoming more valuable than traditional credentials. A Bloomberg study of 500 London hiring managers found that 94 percent would consider candidates without finance backgrounds if they had strong AI and coding skills. Bloomberg also launched BQuant Assistant, a conversational AI coding tool designed for financial analysis. Meanwhile, WEX executive Taranjit Singh warns that faster payments and embedded finance are creating new security risks, especially when AI fraud detection is not paired with strong data quality and clear rules.
On the cybersecurity front, experts say AI safety testing is not keeping pace with jailbreak attacks, where adversaries bypass safety guardrails. Manual testing is too slow and inconsistent, leading to calls for continuous, standardized testing methods similar to software regression testing. At the same time, NIST is developing a framework for AI-powered cybersecurity that aims to improve threat detection and online safety.
Tampa has moved to halt new AI data center construction after the City Council advanced a proposal by Chair Alan Clendenin. The pause follows concerns from Hillsborough County about the water and electricity demands these facilities place on local infrastructure. Governor Ron DeSantis has already signed related legislation in the state.
In education, Professor Prathosh AP from IISc Bangalore is urging students to study mathematics before diving into AI. He recommends a three-semester path beginning with math, then basic machine learning, then advanced methods. He compares using AI without understanding the math to driving a car without knowing how the engine works.
Qodo has introduced a new review layer that maps connections across a codebase to catch issues before they reach the pull request stage. However, the company has not shared specific data on how well the tool performs, and experts say its effectiveness remains unproven until published evidence is available.
Key Takeaways
- Pennsylvania has no law banning AI deepfakes in political ads, though a consumer disclosure law takes effect in December after Stacy Garrity released an AI-generated ad featuring a fake Josh Shapiro.
- Anthropic CEO Dario Amodei revealed the company has run a molecular biology lab since spring 2026, discovering a new enzyme system using AI, raising oversight concerns.
- Electricity use from major tech firms like Microsoft and Google doubled between 2017 and 2021, but emissions could be cut by up to 99 percent through smaller models and green energy optimization.
- A Bloomberg study of 500 London hiring managers found 94 percent would consider candidates without finance backgrounds if they had strong AI and coding skills.
- Tampa City Council advanced a proposal to pause new AI data center construction due to water and electricity demands on local infrastructure.
- Experts recommend treating AI security testing like continuous regression testing to better handle jailbreak attacks at scale.
- IISc Professor Prathosh AP advises students to study math for three semesters before learning AI to build deeper understanding and clearer thinking.
- Qodo introduced a graph-aware code review layer, but its effectiveness remains unproven without published accuracy data.
- WEX warns that payment security is only as strong as the weakest software vendor in the supply chain when AI and embedded finance are involved.
- NIST is developing a framework for AI-powered cybersecurity to improve detection of malicious activity and phishing attacks.
Pennsylvania campaigns use AI-generated ads drawing public scrutiny
Political ads in Pennsylvania now use artificial intelligence to create fake images and voices of candidates. Republican Stacy Garrity used an AI version of Governor Josh Shapiro in a campaign ad. The ad did not clearly tell viewers it was AI generated until late in the video. Experts and lawmakers are debating rules for AI in political ads, and a new law requiring disclosure in consumer ads will take effect in December.
Pennsylvania lacks rules for AI deepfakes in political ads
A recent AI generated ad by Republican Stacy Garrity shows a fake version of Democrat Josh Shapiro speaking and acting in ways he never did. The ad does not warn viewers that it is artificial until about 45 seconds in. Pennsylvania currently has no law that bans deepfakes in political ads or requires disclaimers. Lawmakers are discussing possible changes, but the midterm elections are close.
Anthropic's AI biology lab raises oversight concerns
Anthropic CEO Dario Amodei announced that the company has been running a molecular biology lab in the Bay Area since spring 2026. The lab discovered a new enzyme system using AI. An opinion writer warns that AI tools are hard to control and questions why companies with weak digital safeguards can run physical labs. The article calls for stronger oversight of industry led AI research compared to university labs.
IISc professor says study math to understand AI
Professor Prathosh AP from IISc Bangalore says students should learn mathematics before studying artificial intelligence. He advises a three semester path starting with math, then basic machine learning, then advanced methods. He compares using AI to driving a car without knowing how the engine works. He says deeper math skills help people create new algorithms and think more clearly about technology.
AI and embedded finance increase payment security challenges
WEX executive Taranjit Singh says faster payments and embedded finance create new risks for businesses. He explains that AI can help detect fraud but must be paired with good data and clear rules. He warns that payment security is only as strong as the weakest software vendor in the supply chain. The article stresses the need for visibility, reconciliation, and strong risk controls in modern payment systems.
Making AI Greener: Reducing Tech's Carbon Footprint
AI and tech companies like Microsoft and Google are causing big increases in carbon emissions. Electricity use from major tech firms doubled between 2017 and 2021. Charles Humble suggests using smaller models, running workloads in green energy locations, and optimizing job execution to cut emissions by up to 99 percent.
AI Safety Testing Falls Short Against Jailbreak Attacks
AI systems face growing security risks from jailbreaks where attackers bypass safety guardrails. Manual testing is too slow and inconsistent to handle these threats at scale. Experts recommend treating AI security like continuous regression testing with repeatable, standardized methods. This approach provides better results for compliance and long-term protection.
AI Skills Now Key to Quant Investing Careers
A Bloomberg study of 500 London financial professionals found that 94 percent of hiring managers would consider candidates without traditional finance backgrounds if they had strong AI and coding skills. AI fluency ranked higher than advanced degrees for breaking into quant investing. Bloomberg also launched BQuant Assistant, a conversational AI coding tool for financial analysis.
Tampa Moves to Halt New AI Data Center Construction
Tampa City Council advanced a proposal by Chair Alan Clendenin to pause new AI data center development. The motion directs staff to stop accepting applications while regulations are drafted. This follows Governor Ron DeSantis signing related legislation and Hillsborough County concerns about water and electricity demands from such facilities.
AI Offers Hope for a Safer Online World
New developments in AI are helping improve cybersecurity by detecting and blocking malicious activity and phishing attacks. NIST is developing a framework for AI-powered cybersecurity with guidelines and best practices. While cybersecurity remains challenging, AI is seen as a major step forward in creating a safer internet.
Qodo warns one code change can break software stack
Daniel Singer reported that a single line of code can break a software system even if standard checks pass. Gannaway from Qodo explained that code generation is speeding up faster than human reviews can keep up. He introduced a new review layer that uses a software graph to map connections between different parts of a codebase. This tool aims to catch issues like public API changes before they reach the pull request stage. However, Gannaway did not share specific data on how well the tool works or how many errors it catches. Experts note that until Qodo publishes proof of its accuracy, the effectiveness of this graph-aware review remains unproven.
Sources
- Use of artificial intelligence in campaign ads draws scrutiny
- Pa. law doesn't regulate AI-generated candidates in election ads, but some hope to change that
- Hosseini: Anthropic’s molecular biology lab shows why AI-directed research needs oversight
- ‘Study math if you want to understand AI’: IISc Bangalore professor explains how to learn artificial intelligence
- AI and Embedded Finance Raise the Bar for Payment Controls
- How to Make Artificial Intelligence More Environmentally Sustainable
- Your AI Passed Its Safety Test. It Shouldn't Have
- Bloomberg: AI skills open doors to quant investing
- Tampa City Council advances Alan Clendenin's AI data center moratorium proposal
- A Safer Internet, Thanks to AI
- One-Line Change Can Break the Stack, Qodo Warns
Comments
Please log in to post a comment.