OpenAI is gearing up to release an AI system that’s truly “open,” meaning it’ll be available for download at no cost and not gated behind an API. Beyond its benchmark performance, OpenAI may have a key feature up its sleeve — one that could make its open “reasoning” model highly competitive. Company leaders have been discussing plans to enable the open model to connect to OpenAI’s cloud-hosted models to better answer complex queries. OpenAI CEO Sam Altman described the capability as a “handoff.” If the feature — as sources describe it — makes it into the open model, it will be able to make calls to the OpenAI API to access the company’s other, larger models for a substantial computational lift. It’s unclear if the open model will have the ability to access some of the many tools OpenAI’s models can use, like web search and image generation. The idea for the handoff feature was suggested by a developer during one of OpenAI’s recent developer forums, according to a source. The suggestion appears to have gained traction within the company. OpenAI has been hosting a series of community feedback events with developers to help shape its upcoming open model release. A local model that can tap into more powerful cloud systems brings to mind Apple Intelligence, Apple’s suite of AI capabilities that uses a combination of on-device models and models running in “private” data centers. OpenAI stands to benefit in obvious ways. Beyond generating incremental revenue, a handoff could rope more members of the open source community into the company’s premium ecosystem.
Amazon’s new benchmark to evaluate AI coding agents’ ability to navigate and understand complex codebases and GitHub issues
Amazon has introduced SWE-PolyBench, the first industry benchmark to evaluate AI coding agents’ ability to navigate and understand complex codebases. The benchmark, which measures system performance in GitHub issues, has spurred the development of capable coding agents and has become the de-facto standard for coding agent benchmarking. SWE-PolyBench contains over 2,000 curated issues in four languages and a stratified subset of 500 issues for rapid experimentation. The benchmark aims to advance AI performance in real-world scenarios. Key features of SWE-PolyBench at a glance: Multi-Language Support: Java (165 tasks), JavaScript (1017 tasks), TypeScript (729 tasks), and Python (199 tasks). Extensive Dataset: 2110 instances from 21 repositories ranging from web frameworks to code editors and ML tools, on the same scale as SWE-Bench full with more repository. Task Variety: Includes bug fixes, feature requests, and code refactoring. Faster Experimentation: SWE-PolyBench500 is a stratified subset for efficient experimentation. Leaderboard: A leaderboard with a rich set of metrics for transparent benchmarking.
VoPay’s solution enables banks and SaaS platforms to offer full-stack embedded real-time cross border payments by using APIs, white-label deployment or through no-code services
Embedded financial technology provider VoPay has announced the launch of its new Cross-Border Payments-As-A-Service solution. The white-label technology will enable organisations to move money around the world with full transparency, compliance, and real-time FX. In addition, being built for software platforms and financial institutions, the product will allow firms to benefit from its features without the need to build their own cross-border infrastructure. With this solution, partners and collaborators will have the possibility to expand into global markets and deliver a secure cross-border payment offering. VoPay will continue to focus on meeting the needs, preferences, and demands of clients and users in an ever-evolving market, while prioritising the process of remaining compliant with the regulatory requirements and laws of the industry as well. In addition, VoPay’s solution is purpose-built to serve vertical SaaS platforms, digital marketplaces, and ERP providers, as well as payroll platforms and financial institutions that are looking to unlock new revenue from international money movement. Included in its key capabilities are a full-stack embedded payments, real-time FX engine, compliance by design, end-to-end orchestration, white-label flexibility, and speed to market features. Software platforms and FIs can launch Cross-Border Payments in a way that fits their business model. This includes the possibility to embed the experience directly into their platform using branded UI components and developer-first APIs, launch a Turnkey Solution using VoPay’s white-label deployment model, or access no-code services from the company’s dashboard with full functionality–zero development needed.
Bilt Rewards partners with Southwest Airlines to enable members to turn points earned from housing payments into points that can be used to make flight bookings
Bilt Rewards has partnered with Southwest Airlines to enable Bilt members to turn points earned from housing payments into points that can be used to book flights. Bilt members can convert their Bilt Points into Southwest Rapid Rewards points at a 1-to-1 ratio. Bilt Points are earned each time members pay their rent on time through Bilt. “Now, your housing payment that was once just a monthly expense can unlock travel experiences with one of America’s most popular airlines, giving our members access to more destinations across the U.S. and beyond,” Bilt Founder and CEO Ankur Jain said. Bilt and Southwest plan to add exclusive promotions and enhanced capabilities for their customers as their new partnership evolves. Jonathan Clarkson, vice president and chief product officer at Southwest Airlines, said that the collaboration “allows Southwest to connect with a unique audience of renters and homeowners across the United States while offering them the opportunity to use their Bilt Points to fly Southwest.” Bilt Points can now be used toward travel across over 100 airlines and hotel partners, fitness classes at boutique studios, art and home decor through the Bilt Home Design Collection, rent credits, future down payments on a home, or eligible student loans.
Two’s B2B BNPL solution can be embedded directly into merchants’ checkouts, providing instant trade credit at the POS enabling consolidating multiple purchases into grouped monthly statements
A new wave of European startups is actively pursuing expansion into the US B2B BNPL sector as part of broader international growth strategies. Among them is Two, a B2B BNPL firm currently running a pilot program in North America as it prepares for a full-scale launch into the region’s BNPL market. While Two doesn’t yet have a physical footprint in the United States, it maintains an online presence through partnerships with Santander and Allianz, whose operations span the Americas, including the US. Two’s current pilot program offers trade credit solutions through its white-labeled Buy Now, Pay Later service and its installment product. These solutions are integrated into merchants’ checkout, existing financial workflows, and payment systems, enabling buyers to break large purchases over up to 24 months. Andreas Mjelde, CEO & Co-Founder of Two, shares that the company is preparing to roll out its full product lineup in North America by the second quarter of 2025. The full lineup will include: 1) B2B BNPL embedded directly into guest checkouts, providing instant trade credit to businesses at the point of sale. Currently, BNPL is embedded in certain open and closed-wall checkouts. 2) Trade Accounts that will enable businesses to consolidate multiple purchases into grouped monthly statements, simplifying payment management and cash flow oversight. 3) Extended installment plans, providing repayment options of up to 36 months for businesses managing larger transactions. He explains that the firm’s localized approach incorporates regional credit risk assessments and personalized repayment experience targeting the specific needs of US-based companies. ”Our proprietary risk models, Frida and Delphi, deliver high acceptance rates while minimizing friction for buyers,” says Andreas. This automation speeds up onboarding and boosts approval rates, without the need for hard credit checks. The firm has built its machine learning infrastructure and AI models in-house. Two is pursuing a phased expansion strategy, starting in US states with high digital payment adoption with the support of global banking partnerships, before scaling further
Ibanera’s onboarding and AML compliance solution adopts a layered and tokenized design that replaces sensitive data with unique identifiers and makes compliance traceable and verifiable across complex networks of users
Digital banking platform Ibanera announced the launch of its ‘Nested Compliance’. Produced by the company CEO, Michael Carbonara, the infrastructure directly combats current risks faced by the Fintech and Banking as a Service (BaaS) industry. The framework differs from current financial compliance processes by automating KYC and AML processes while adopting a layered and tokenized design that embeds compliance across its vast customer network. Ibanera’s new Nested Compliance concept introduces high-level automation for KYC and AML processes for operational efficiency. The upgraded system uses API-driven, real-time compliance monitoring with early warning triggers for expiring KYC, automated counterparty verification, and embedded reporting to speed up processes and keep detailed records for audit purposes. The framework also debuts a unique design that tokenizes compliance. Sensitive data is replaced with unique identifiers, minimizing PII (Personally Identifiable Information) exposure risk. The tokenized design also creates a system where compliance is traceable and verifiable across complex networks of users, simplifying financial relationships and the overall compliance process. Through this, layered transactions are secure, monitored, and compliant with FATF, FINCEN, and global Anti-Money Laundering (AML) regulations. The new Nested Compliance protocol is being rolled out on Ibanera’s ecosystem this year and is available for customers and partners. Key updates include real-time risk assessments of counterparty data, along with ongoing AML checks to prevent illicit activities within nested payment structures.
P2P payment information network Phixius by Nacha partners Kinexys by J.P. Morgan to add near real-time global validation of bank account ownership, status and transactions to its real-time validation
Phixius by Nacha and Kinexys Liink established information exchanges between their payment information networks. The collaboration will enhance account validation coverage for financial institutions, FinTechs and corporations. Phixius, which is a peer-to-peer payment information network, will serve as Kinexys Liink’s key U.S. payment information network responder, enabling near real-time validation of domestic bank account data. Kinexys Liink, which is a bank-led peer-to-peer data-sharing network and part of Kinexys by J.P. Morgan, will allow Phixius participants access to its Confirm application’s global account validation capabilities, expanding the reach of the Phixius network. “Kinexys Liink and Phixius customers can benefit by validation services using data provided by either network, helping to mitigate payment fraud and reduce potential ACH returns,” Rob Unger, managing director of ACH Network development at Nacha, said. Gloria Wan, general manager of Kinexys Liink at Kinexys by J.P. Morgan, said: “Through the collaboration with Phixius by Nacha, we look forward to expanding the reach of Kinexys Liink to further strengthen account validation and cross-border payment infrastructure globally.”
Microsoft releases taxonomy of failure modes- security and safety- inherent to agentic architecture- novel modes unique to agentic systems (e.g. agent compromise) and modes representing amplification of existing GenAI risks (e.g. bias amplification)
Microsoft’s AI Red Team has published a detailed taxonomy addressing the failure modes inherent to agentic architectures. Agentic AI systems are autonomous entities that observe and act upon their environment to achieve predefined objectives. These systems integrate capabilities such as autonomy, environment observation, interaction, memory, and collaboration. However, these features introduce a broader attack surface and new safety concerns. The report distinguishes between novel failure modes unique to agentic systems and amplification of risks already observed in generative AI contexts. Microsoft categorizes failure modes across security and safety dimensions. Novel Security Failures: Including agent compromise, agent injection, agent impersonation, agent flow manipulation, and multi-agent jailbreaks. Novel Safety Failures: Covering issues such as intra-agent Responsible AI (RAI) concerns, biases in resource allocation among multiple users, organizational knowledge degradation, and prioritization risks impacting user safety. Existing Security Failures: Encompassing memory poisoning, cross-domain prompt injection (XPIA), human-in-the-loop bypass vulnerabilities, incorrect permissions management, and insufficient isolation. Existing Safety Failures: Highlighting risks like bias amplification, hallucinations, misinterpretation of instructions, and a lack of sufficient transparency for meaningful user consent
FDX standard APIs adoption hits 114m customer connections , a 50% increase from the comparable figure of 76 million a year ago helping companies to “improve interoperability when they integrate”
FDX has reported that approximately 114 million customer connections are now happening through APIs aligned to the FDX standard, as the use of standardized APIs to enable consumer-permission data sharing continues to grow. The growth in FDX API adoption to 114 million customer connections represents a 50% increase from the comparable figure of 76 million a year ago, while FDX called it “a sizable jump” from the 96 million reported six months ago. “Hitting 114 million customers connections reflects both the scale of FDX’s impact and the power of industry-led collaboration to drive interoperability,” said Kevin Feltes, chief executive officer of FDX. The organization stated that the FDX API standard “is solidly rooted as the leading method in North America for building APIs to enable safe, user-permissioned connections” and said that “industry-led standard-setting and collaboration” is helping companies to “improve interoperability when they integrate”. FDX noted that API adoption continued to grow “amid a shifting regulatory environment”, but pointed to the “work ahead”. According to FDX, “tens of millions of consumers and small businesses in North America are still sharing financial data through methods that require sharing login credentials with third parties and may offer less customer control”.
CrowdStrike’s SIEM solution is first to bring managed threat hunting to third-party data- unifies real-time intelligence and AI-driven automation to deliver expert-led threat hunting across all attack surfaces
CrowdStrike introduced Falcon® Adversary OverWatch Next-Gen SIEM, the first and only solution to bring managed threat hunting to third-party data. This breakthrough innovation extends the visibility of CrowdStrike’s elite threat hunters into unmanaged attack surfaces adversaries have long exploited. By leveraging third-party data ingested by Falcon® Next-Gen SIEM, CrowdStrike delivers 24/7 expert detection beyond endpoints, identity and cloud environments to stop breaches across every attack surface. Powered by the AI-native CrowdStrike Falcon® cybersecurity platform, Falcon Adversary OverWatch uses deep adversary expertise and industry-leading threat intelligence to rapidly uncover evasive threats. Falcon Next-Gen SIEM unifies native and third-party data, real-time intelligence and AI-driven automation to deliver comprehensive visibility, high-fidelity alerts and machine speed response. New innovations include: Expert-Led Threat Hunting Across all Attack Surfaces; UEBA and Case Management for Falcon Next-Gen SIEM; Unified Identity Security and Next-Gen SIEM; CrowdStrike Pulse Services.