Connect with us

AI Research

Senator Reintroduces Bill for AI Regulatory Sandbox for Finance

Published

1 month ago

on


A bipartisan bill to establish regulatory sandboxes for artificial intelligence (AI) experimentation in financial services took center stage at a Senate subcommittee hearing Wednesday (July 30), as lawmakers weighed how to balance AI-driven innovation with oversight.

Sen. Mike Rounds (R-S.D.), who chairs the Senate Subcommittee on Securities, Insurance, and Investment, announced the reintroduction of the “Unleashing AI Innovation in Financial Services Act.”

The bill, co-sponsored with Sen. Martin Heinrich, D-N.M., would let financial institutions test AI-enabled products and services without immediate risk of enforcement action, as long as they meet transparency, consumer protection and national security requirements.

“By creating a safe space for experimentation, we can help firms innovate and regulators learn without applying outdated rules that don’t fit today’s technology,” Rounds said. The bill was originally introduced in 2024.

If enacted, S.4951 would direct financial regulators — including the Securities and Exchange Commission, Consumer Financial Protection Bureau and the Federal Reserve — to evaluate and potentially waive or modify existing rules for approved AI test projects. Agencies would have 90 days to approve or deny applications, with automatic approval if no decision is made by the deadline.

During the hearing, lawmakers from both parties said they wished to foster innovation while mitigating the risks of unregulated AI adoption.

Sen. Mark Warner, D-Va., remembered a prior hearing called by Sen. Chuck Schumer, D-N.Y., that brought together CEOs of top AI companies. “Remember Schumer asked, ‘How many of you all think AI needs to be regulated?’ Everybody raised their hand.”

But once it came down to brass tacks, “I worry that we’re frankly going almost completely in the opposite direction,” Warner said. For example, President Trump’s AI action plan favored deregulation of AI.

Warner said that if people could turn back time, “most of us would think that if in 2014 we’d put some guardrails on social media, at least [to protect] our kids’ mental health, we’d be in a better spot. We didn’t — and social media is tiny compared to the potential that AI has.”

Warner pointed to the example of Delta Air Lines testing AI systems that use an individual’s data to tailor airfare pricing, a practice he called “surveillance” pricing. Warner and two other senators are concerned and have written a letter to Delta asking for additional information.

Read more: Delta Air Lines Tests AI-Powered Personalized Pricing

Insurers Refuse to Cover AI Risks

Kevin Kalinich, global leader for intangible assets at Aon, said during the hearing that the insurance industry is beginning to respond to the risks posed by emerging AI capabilities, including hallucinations from generative models, deepfakes, and autonomous software agents.

However, Kalinich said that actuarial models lack sufficient historical data to accurately price these risks. As a result, some insurers are excluding AI-related exposures in professional liability and cyber policies.

Meanwhile, “a few cutting-edge insurance carriers have created AI-specific insurance protection, albeit with smaller limits than are sufficient for larger clients,” Kalinich said.

The Aon executive noted that underwriters are more likely to offer favorable terms when firms have strong AI governance practices, including documented model audits, explainability metrics and bias mitigation protocols. “Good governance leads to better insurability, which in turn supports innovation and consumer protection,” he said.

Tal Cohen, president of Nasdaq, described how AI is already improving market surveillance, reducing false positives and streamlining investigations. Last week, Nasdaq launched its agentic AI workforce for compliance and efficiency.

Nasdaq’s first two AI agents — the digital sanctions analyst and digital enhanced due diligence analyst — were put to work to labor-intensive compliance tasks. For the digital sanctions analyst, when integrated into a bank’s alert triage workflow, reduced the review workload by over 80%.

Beyond efficiency is stability. Rounds asked Nasdaq’s Cohen what threats from adversarial nations might be coming that would destabilize U.S. financial markets, since delays of even milliseconds in order execution can erode investor confidence.

Cohen said that Nasdaq’s chief information security officer not only uses the most advanced AI cybersecurity tools but also coordinates with industry peers on protection. “This is not a competitive element for us with other exchanges,” he said. “We share and we collaborate.”

But when pressed whether a formal multiagency task force exists to address AI risks across exchanges, Cohen replied, “We need one. We need to have that discussion.”

Moreover, the liability arising from these AI incidents would be “shared,” Cohen added.

David Cox, vice president for AI models at IBM Research, said an open-source approach in AI development is important in building trust.

 “We strongly believe in the value of open source AI. It enhances security, trust and collaboration through transparency, enables smaller firms and research organizations to compete without prohibitive upfront capital costs, and it expands the pipeline of future talent,” Cox said.

Large language models (LLMs) must be auditable, particularly in regulated environments.

“Firms must understand exactly what data underpins their models and be able to audit those systems over time,” Cox said, adding that few model developers disclose their training datasets, making compliance tougher.

Sen. Katie Britt, R-Ala., raised concerns about AI-powered impersonation scams, citing a 148% year-over-year increase in financial fraud driven by generative AI. She also asked Cohen about trading bots and the risk of AI-based decision systems on market integrity. Cohen said any regulated firm would have the “proper controls” in place.

In the end, there was broad agreement at the hearing that the status quo on regulations is not sufficient. Warner noted that China is no longer merely copying technology instead of innovating. “That’s changed,” he said. “China is not playing for second place in the race for AI.”

Added Britt: “This is the race that matters.”

Read more:

California Advances Bill Regulating AI Companions Amid Concerns Over Mental Health Issues

AI Regulations Bring Deluge of Lobbying Efforts to DC

Senate Shoots Down 10-Year Ban on State AI Regulations

 

 



Source link

Related Topics:
Continue Reading
Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *

AI Research

MAIA platform for routine clinical testing: an artificial intelligence embryo selection tool developed to assist embryologists

Published

1 hour ago

on

September 1, 2025

By


  • Graham, M. E. et al. Assisted reproductive technology: Short- and long-term outcomes. Dev. Med. Child. Neurol. 65, 38–49 (2023).

    PubMed 

    Google Scholar     

  • Jiang, V. S. & Bormann, C. L. Artificial intelligence in the in vitro fertilization laboratory: a review of advancements over the last decade. Fertil. Steril. 120, 17–23 (2023).

    PubMed 

    Google Scholar     

  • Devine, K. et al. Single vitrified blastocyst transfer maximizes liveborn children per embryo while minimizing preterm birth. Fertil. Steril. 103, 1454–1460 (2015).

    PubMed 
    PubMed Central 

    Google Scholar     

  • Tiitinen, A. Single embryo transfer: why and how to identify the embryo with the best developmental potential. Best Pract. Res. Clin. Endocrinol. Metab. 33, 77–88 (2019).

    PubMed 

    Google Scholar     

  • Glatstein, I., Chavez-Badiola, A. & Curchoe, C. L. New frontiers in embryo selection. J. Assist. Reprod. Genet. 40, 223–234 (2023).

    PubMed 
    PubMed Central 

    Google Scholar     

  • Gardner, D. K. & Schoolcraft, W. B. Culture and transfer of human blastocysts. Curr. Opin. Obstet. Gynaecol. 11, 307–311 (1999).


    Google Scholar     

  • Sciorio, R. & Meseguer, M. Focus on time-lapse analysis: blastocyst collapse and morphometric assessment as new features of embryo viability. Reprod. BioMed. Online. 43, 821–832 (2021).

    PubMed 

    Google Scholar     

  • Sundvall, L., Ingerslev, H. J., Knudsen, U. B. & Kirkegaard, K. Inter- and intra-observer variability of time-lapse annotations. Hum. Reprod. 28, 3215–3221 (2013).

    PubMed 

    Google Scholar     

  • Gallego, R. D., Remohí, J. & Meseguer, M. Time-lapse imaging: the state of the Art. Biol. Reprod. 101, 1146–1154 (2019).

    PubMed 

    Google Scholar     

  • VerMilyea, M. D. et al. Computer-automated time-lapse analysis results correlate with embryo implantation and clinical pregnancy: a blinded, multi-centre study. Reprod. Biomed. Online. 29, 729–736 (2014).

    PubMed 
    PubMed Central 

    Google Scholar     

  • Chéles, D. S., Molin, E. A. D., Rocha, J. C. & Nogueira, M. F. G. Mining of variables from embryo morphokinetics, blastocyst’s morphology and patient parameters: an approach to predict the live birth in the assisted reproduction service. JBRA Assist. Reprod. 24, 470–479 (2020).

    PubMed 
    PubMed Central 

    Google Scholar     

  • Rocha, C., Nogueira, M. G., Zaninovic, N. & Hickman, C. Is AI assessment of morphokinetic data and digital image analysis from time-lapse culture predictive of implantation potential of human embryos? Fertil. Steril. 110, e373 (2018).


    Google Scholar     

  • Zaninovic, N. et al. Application of artificial intelligence technology to increase the efficacy of embryo selection and prediction of live birth using human blastocysts cultured in a time-lapse incubator. Fertil. Steril. 110, e372–e373 (2018).


    Google Scholar     

  • Alegre, L. et al. First application of artificial neuronal networks for human live birth prediction on Geri time-lapse monitoring system blastocyst images. Fertil. Steril. 114, e140 (2020).


    Google Scholar     

  • Bori, L. et al. An artificial intelligence model based on the proteomic profile of euploid embryos and blastocyst morphology: a preliminary study. Reprod. BioMed. Online. 42, 340–350 (2021).

    PubMed 

    Google Scholar     

  • Chéles, D. S. et al. An image processing protocol to extract variables predictive of human embryo fitness for assisted reproduction. Appl. Sci. 12, 3531 (2022).


    Google Scholar     

  • Jacobs, C. K. et al. Embryologists versus artificial intelligence: predicting clinical pregnancy out of a transferred embryo who performs it better? Fertil. Steril. 118, e81–e82 (2022).


    Google Scholar     

  • Lorenzon, A. et al. P-211 development of an artificial intelligence software with consistent laboratory data from a single IVF center: performance of a new interface to predict clinical pregnancy. Hum. Reprod. 39, deae108.581 (2024).

  • Fernandez, E. I. et al. Artificial intelligence in the IVF laboratory: overview through the application of different types of algorithms for the classification of reproductive data. J. Assist. Reprod. Genet. 37, 2359–2376 (2020).

    PubMed 
    PubMed Central 

    Google Scholar     

  • Mendizabal-Ruiz, G. et al. Computer software (SiD) assisted real-time single sperm selection associated with fertilization and blastocyst formation. Reprod. BioMed. Online. 45, 703–711 (2022).

    PubMed 

    Google Scholar     

  • Fjeldstad, J. et al. Segmentation of mature human oocytes provides interpretable and improved blastocyst outcome predictions by a machine learning model. Sci. Rep. 14, 10569 (2024).

    ADS 
    PubMed 
    PubMed Central 

    Google Scholar     

  • Khosravi, P. et al. Deep learning enables robust assessment and selection of human blastocysts after in vitro fertilization. NPJ Digit. Med. 2, 21 (2019).

    PubMed 
    PubMed Central 

    Google Scholar     

  • Hickman, C. et al. Inner cell mass surface area automatically detected using Chloe eq™(fairtility), an ai-based embryology support tool, is associated with embryo grading, embryo ranking, ploidy and live birth outcome. Fertil. Steril. 118, e79 (2022).


    Google Scholar     

  • Tran, D., Cooke, S., Illingworth, P. J. & Gardner, D. K. Deep learning as a predictive tool for fetal heart pregnancy following time-lapse incubation and blastocyst transfer. Hum. Reprod. 34, 1011–1018 (2019).

    PubMed 
    PubMed Central 

    Google Scholar     

  • Rajendran, S. et al. Automatic ploidy prediction and quality assessment of human blastocysts using time-lapse imaging. Nat. Commun. 15, 7756 (2024).

    PubMed 
    PubMed Central 

    Google Scholar     

  • Bormann, C. L. et al. Consistency and objectivity of automated embryo assessments using deep neural networks. Fertil. Steril. 113, 781–787e1 (2020).

    PubMed 
    PubMed Central 

    Google Scholar     

  • Kragh, M. F. & Karstoft, H. Embryo selection with artificial intelligence: how to evaluate and compare methods? J. Assist. Reprod. Genet. 38, 1675–1689 (2021).

    PubMed 
    PubMed Central 

    Google Scholar     

  • Cromack, S. C., Lew, A. M., Bazzetta, S. E., Xu, S. & Walter, J. R. The perception of artificial intelligence and infertility care among patients undergoing fertility treatment. J. Assist. Reprod. Genet. https://doi.org/10.1007/s10815-024-03382-5 (2025).

    PubMed 

    Google Scholar     

  • Fröhlich, H. et al. From hype to reality: data science enabling personalized medicine. BMC Med. 16, 150 (2018).

    PubMed 
    PubMed Central 

    Google Scholar     

  • Zhu, J. et al. External validation of a model for selecting day 3 embryos for transfer based upon deep learning and time-lapse imaging. Reprod. BioMed. Online. 47, 103242 (2023).

    PubMed 

    Google Scholar     

  • Yelke, H. K. et al. O-007 Simplifying the complexity of time-lapse decisions with AI: CHLOE (Fairtility) can automatically annotate morphokinetics and predict blastulation (at 30hpi), pregnancy and ongoing clinical pregnancy. Hum. Reprod. 37, deac104.007 (2022).

  • Papatheodorou, A. et al. Clinical and practical validation of an end-to-end artificial intelligence (AI)-driven fertility management platform in a real-world clinical setting. Reprod. BioMed. Online. 45, e44–e45 (2022).


    Google Scholar     

  • Salih, M. et al. Embryo selection through artificial intelligence versus embryologists: a systematic review. Hum. Reprod. Open hoad031 (2023).

  • Nunes, K. et al. Admixture’s impact on Brazilian population evolution and health. Science. 388(6748), eadl3564 (2025).

  • Jackson-Bey, T. et al. Systematic review of Racial and ethnic disparities in reproductive endocrinology and infertility: where do we stand today? F&S Reviews. 2, 169–188 (2021).


    Google Scholar     

  • Kassi, L. A. et al. Body mass index, not race, May be associated with an alteration in early embryo morphokinetics during in vitro fertilization. J. Assist. Reprod. Genet. 38, 3091–3098 (2021).

    PubMed 
    PubMed Central 

    Google Scholar     

  • Pena, S. D. J., Bastos-Rodrigues, L., Pimenta, J. R. & Bydlowski, S. P. DNA tests probe the genomic ancestry of Brazilians. Braz J. Med. Biol. Res. 42, 870–876 (2009).

    PubMed 

    Google Scholar     

  • Fraga, A. M. et al. Establishment of a Brazilian line of human embryonic stem cells in defined medium: implications for cell therapy in an ethnically diverse population. Cell. Transpl. 20, 431–440 (2011).


    Google Scholar     

  • Amin, F. & Mahmoud, M. Confusion matrix in binary classification problems: a step-by-step tutorial. J. Eng. Res. 6, 0–0 (2022).


    Google Scholar     

  • Magdi, Y. et al. Effect of embryo selection based morphokinetics on IVF/ICSI outcomes: evidence from a systematic review and meta-analysis of randomized controlled trials. Arch. Gynecol. Obstet. 300, 1479–1490 (2019).

    PubMed 

    Google Scholar     

  • Guo, Y. H., Liu, Y., Qi, L., Song, W. Y. & Jin, H. X. Can time-lapse incubation and monitoring be beneficial to assisted reproduction technology outcomes? A randomized controlled trial using day 3 double embryo transfer. Front. Physiol. 12, 794601 (2022).

    PubMed 
    PubMed Central 

    Google Scholar     

  • Giménez, C., Conversa, L., Murria, L. & Meseguer, M. Time-lapse imaging: morphokinetic analysis of in vitro fertilization outcomes. Fertil. Steril. 120, 228–227 (2023).


    Google Scholar     

  • Vitrolife EmbryoScope + time-lapse system. (2023). https://www.vitrolife.com/products/time-lapse-systems/embryoscopeplus-time-lapse-system/.

  • Lagalla, C. et al. A quantitative approach to blastocyst quality evaluation: morphometric analysis and related IVF outcomes. J. Assist. Reprod. Genet. 32, 705–712 (2015).

    PubMed 
    PubMed Central 

    Google Scholar     

  • Rocha, J. C. et al. A method based on artificial intelligence to fully automatize the evaluation of bovine blastocyst images. Sci. Rep. 7, 7659 (2017).

    ADS 
    PubMed 
    PubMed Central 

    Google Scholar     

  • Chavez-Badiola, A. et al. Predicting pregnancy test results after embryo transfer by image feature extraction and analysis using machine learning. Sci. Rep. 10, 4394 (2020).

    ADS 
    PubMed 
    PubMed Central 

    Google Scholar     

  • Matos, F. D., Rocha, J. C. & Nogueira, M. F. G. A method using artificial neural networks to morphologically assess mouse blastocyst quality. J. Anim. Sci. Technol. 56, 15 (2014).

    PubMed 
    PubMed Central 

    Google Scholar     

  • Wang, S., Zhou, C., Zhang, D., Chen, L. & Sun, H. A deep learning framework design for automatic blastocyst evaluation with multifocal images. IEEE Access. 9, 18927–18934 (2021).


    Google Scholar     

  • Berntsen, J., Rimestad, J., Lassen, J. T., Tran, D. & Kragh, M. F. Robust and generalizable embryo selection based on artificial intelligence and time-lapse image sequences. PLoS One. 17, e0262661 (2022).

    PubMed 
    PubMed Central 

    Google Scholar     

  • Fruchter-Goldmeier, Y. et al. An artificial intelligence algorithm for automated blastocyst morphometric parameters demonstrates a positive association with implantation potential. Sci. Rep. 13, 14617 (2023).

    ADS 
    PubMed 
    PubMed Central 

    Google Scholar     

  • Illingworth, P. J. et al. Deep learning versus manual morphology-based embryo selection in IVF: a randomized, double-blind noninferiority trial. Nat. Med. 30, 3114–3120 (2024).

    PubMed 
    PubMed Central 

    Google Scholar     

  • Kanakasabapathy, M. K. et al. Development and evaluation of inexpensive automated deep learning-based imaging systems for embryology. Lab. Chip. 19, 4139–4145 (2019).

    PubMed 
    PubMed Central 

    Google Scholar     

  • Loewke, K. et al. Characterization of an artificial intelligence model for ranking static images of blastocyst stage embryos. Fertil. Steril. 117, 528–535 (2022).

    PubMed 

    Google Scholar     

  • Hengstschläger, M. Artificial intelligence as a door opener for a new era of human reproduction. Hum. Reprod. Open hoad043 (2023).

  • Lassen Theilgaard, J., Fly Kragh, M., Rimestad, J., Nygård Johansen, M. & Berntsen, J. Development and validation of deep learning based embryo selection across multiple days of transfer. Sci. Rep. 13 (1), 4235 (2023).

    ADS 

    Google Scholar     

  • Lozano, M. et al. P-301 Assessment of ongoing clinical outcomes prediction of an AI system on retrospective SET data, Human Reprod. 38(Issue Supplement_1), dead093.659. (2023).

  • Collins, G. S. et al. TRIPOD + AI statement: updated guidance for reporting clinical prediction models that use regression or machine learning methods. BMJ 385, e078378 (2024).

    PubMed 
    PubMed Central 

    Google Scholar     

  • Abdolrasol, M. G. M. et al. Artificial neural networks based optimization techniques: a review. Electronics 10, 2689 (2021).


    Google Scholar     

  • Yuzer, E. O. & Bozkurt, A. Instant solar irradiation forecasting for solar power plants using different ANN algorithms and network models. Electr. Eng. 106, 3671–3689 (2024).


    Google Scholar     

  • Guariso, G. & Sangiorgio, M. Improving the performance of multiobjective genetic algorithms: an elitism-based approach. Information 11, 587 (2020).


    Google Scholar     

  • García-Pascual, C. M. et al. Optimized NGS approach for detection of aneuploidies and mosaicism in PGT-A and imbalances in PGT-SR. Genes 11, 724 (2020).

    PubMed 
    PubMed Central 

    Google Scholar     



    • Source link

    Continue Reading

    AI Research

    UCLA Researchers Enable Paralyzed Patients to Control Robots with Thoughts Using AI – CHOSUNBIZ – Chosun Biz

    Published

    4 hours ago

    on

    September 1, 2025

    By



    UCLA Researchers Enable Paralyzed Patients to Control Robots with Thoughts Using AI – CHOSUNBIZ  Chosun Biz



    Source link

    Continue Reading

    AI Research

    Hackers exploit hidden prompts in AI images, researchers warn

    Published

    5 hours ago

    on

    September 1, 2025

    By


    Cybersecurity firm Trail of Bits has revealed a technique that embeds malicious prompts into images processed by large language models (LLMs). The method exploits how AI platforms compress and downscale images for efficiency. While the original files appear harmless, the resizing process introduces visual artifacts that expose concealed instructions, which the model interprets as legitimate user input.

    In tests, the researchers demonstrated that such manipulated images could direct AI systems to perform unauthorized actions. One example showed Google Calendar data being siphoned to an external email address without the user’s knowledge. Platforms affected in the trials included Google’s Gemini CLI, Vertex AI Studio, Google Assistant on Android, and Gemini’s web interface.

    Read More: Meta curbs AI flirty chats, self-harm talk with teens

    The approach builds on earlier academic work from TU Braunschweig in Germany, which identified image scaling as a potential attack surface in machine learning. Trail of Bits expanded on this research, creating “Anamorpher,” an open-source tool that generates malicious images using interpolation techniques such as nearest neighbor, bilinear, and bicubic resampling.

    From the user’s perspective, nothing unusual occurs when such an image is uploaded. Yet behind the scenes, the AI system executes hidden commands alongside normal prompts, raising serious concerns about data security and identity theft. Because multimodal models often integrate with calendars, messaging, and workflow tools, the risks extend into sensitive personal and professional domains.

    Also Read: Nvidia CEO Jensen Huang says AI boom far from over

    Traditional defenses such as firewalls cannot easily detect this type of manipulation. The researchers recommend a combination of layered security, previewing downscaled images, restricting input dimensions, and requiring explicit confirmation for sensitive operations.

    “The strongest defense is to implement secure design patterns and systematic safeguards that limit prompt injection, including multimodal attacks,” the Trail of Bits team concluded.



    Source link

    Continue Reading

    Trending