World Solve

Consumer drones have become cheap and widely available, but affordable detection systems for identifying unauthorized or suspicious drone activity in sensitive areas remain limited. Closing this security gap remains an unresolved technical and regulatory challenge.

Companies sometimes stop providing security patches for older but still commonly used software or devices, leaving users vulnerable without clear warning. Establishing consistent, enforced support timelines remains an unresolved industry practice.

Building one app that works well across many different operating systems and device types often results in compromises that make it less polished than a dedicated, platform-specific version. Fully closing this quality gap remains an unresolved software engineering tradeoff.

Scammers can now use AI to generate convincing images of products that don't exist or don't match what's actually shipped, making it harder for shoppers to identify fraudulent listings. Detection tools on marketplaces remain inconsistently effective.

Attackers can sometimes convince phone carriers to transfer a victim's number to a new device, bypassing two-factor authentication tied to that number. Carriers have improved verification, but this attack method remains an ongoing security gap.

Fraudsters can use AI to generate entirely fake but convincing identities, complete with fabricated financial histories, to open accounts and commit fraud at scale. Detecting these synthetic identities before damage occurs remains a growing, unresolved challenge for financial institutions.

When companies discontinue software or online services, users often lose access to their data or functionality with little warning or support for migrating elsewhere. Standardized, consumer-friendly transition processes remain largely absent from the industry.

Even after an AI model is retired, techniques exist that can sometimes extract fragments of its original training data, posing ongoing privacy risks. Fully preventing this kind of data leakage after decommissioning remains unresolved.

AI hiring systems trained on historical hiring data can inadvertently learn and repeat past patterns of discrimination, even when explicitly designed to avoid it. Fully eliminating this bias while maintaining useful screening capability remains unresolved.

While hardware energy use gets some attention, the carbon footprint of running specific pieces of software, from development to daily use, remains largely unmeasured and unaccounted for. Building this tracking into standard practice remains unresolved.

Companies increasingly build products using AI models developed by others, but there's no consistent, scalable way to verify those models are safe and reliable before integration. This creates hidden risk throughout the growing AI supply chain.

Recommendation algorithms optimize for engagement, which can unintentionally narrow the range of content users are exposed to over time, reinforcing existing beliefs rather than broadening perspective. Designing systems that balance engagement with diversity of exposure remains unresolved.

AI-generated contracts, court filings, or legal notices can appear authentic enough to deceive people unfamiliar with proper legal formatting, creating new fraud risks. Detecting these fakes reliably remains an emerging unresolved challenge.

Without standardized measurement, it's difficult to compare the true environmental cost of different AI systems or hold companies accountable for improving efficiency. A consistent, transparent measurement standard remains undeveloped.

Tools designed to detect AI-written assignments frequently produce false positives, wrongly flagging legitimate student work as AI-generated. This unreliability creates real consequences for students while genuine cheating sometimes slips through undetected.

Fake reviews generated at scale by AI can be nearly indistinguishable from genuine customer feedback, undermining trust in online shopping decisions. Detection tools are struggling to keep pace with how convincing these fakes have become.

It's often unclear exactly what data went into training a given AI model, making it difficult to audit for bias, copyright issues, or harmful content. Greater transparency into training data remains technically and legally unresolved.

Advanced spyware can be installed on phones without any visible sign, and even security experts sometimes struggle to detect its presence through standard tools. Better, more accessible detection methods remain an urgent unmet need.

Researchers have shown that inaudible or disguised audio signals can sometimes trigger voice assistants to take unintended actions without the user's knowledge. Fully closing this security gap remains unresolved.

AI health tools can sometimes provide advice that sounds plausible but is subtly wrong or dangerous, and there's no consistent, scalable system for catching these errors before they reach real patients. This gap grows more urgent as AI health tools become more widely used.