virtual-insanity
← 뒤로

260314 tg (9개)

seedling aggregate 2026-03-14

260314 Telegram 모음

[6/10] Security Considerations for Artificial Intelligence Agents This article, a light

Security Considerations for Artificial Intelligence Agents This article, a lightly adapted version of Perplexity's response to NIST/CAISI Request for Information 2025-0035, details our observations and recommendations concerning the security of frontier AI agents. These insights are informed by Perplexity's experience operating general-purpose agentic systems used by millions of users and thousands of enterprises in both controlled and open-world environments. Agent architectures change core assumptions around code-data separation, authority boundaries, agent_orchestration ron agent multi-agent

점수: 6/10 — 점수 6/10: multi-agent


[6/10] Deep Incentive Design with Differentiable Equilibrium Blocks Automated design of

Deep Incentive Design with Differentiable Equilibrium Blocks Automated design of multi-agent interactions with desirable equilibrium outcomes is inherently difficult due to the computational hardness, non-uniqueness, and instability of the resulting equilibria. In this work, we propose the use of game-agnostic differentiable equilibrium blocks (DEBs) as modules in a novel, differentiable framework to address a wide variety of incentive design problems from economics and computer science. We call this framework deep incentive design (DID). To validate our agent_orchestration agent multi-agent

점수: 6/10 — 점수 6/10: multi-agent


[6/10] Security Considerations for Artificial Intelligence Agents This article, a light

Security Considerations for Artificial Intelligence Agents This article, a lightly adapted version of Perplexity's response to NIST/CAISI Request for Information 2025-0035, details our observations and recommendations concerning the security of frontier AI agents. These insights are informed by Perplexity's experience operating general-purpose agentic systems used by millions of users and thousands of enterprises in both controlled and open-world environments. Agent architectures change core assumptions around code-data separation, authority boundaries, agent_orchestration ron agent multi-agent

점수: 6/10 — 점수 6/10: multi-agent


[6/10] Deep Incentive Design with Differentiable Equilibrium Blocks Automated design of

Deep Incentive Design with Differentiable Equilibrium Blocks Automated design of multi-agent interactions with desirable equilibrium outcomes is inherently difficult due to the computational hardness, non-uniqueness, and instability of the resulting equilibria. In this work, we propose the use of game-agnostic differentiable equilibrium blocks (DEBs) as modules in a novel, differentiable framework to address a wide variety of incentive design problems from economics and computer science. We call this framework deep incentive design (DID). To validate our agent_orchestration agent multi-agent

점수: 6/10 — 점수 6/10: multi-agent


[6/10] Security Considerations for Artificial Intelligence Agents This article, a light

Security Considerations for Artificial Intelligence Agents This article, a lightly adapted version of Perplexity's response to NIST/CAISI Request for Information 2025-0035, details our observations and recommendations concerning the security of frontier AI agents. These insights are informed by Perplexity's experience operating general-purpose agentic systems used by millions of users and thousands of enterprises in both controlled and open-world environments. Agent architectures change core assumptions around code-data separation, authority boundaries, agent_orchestration ron agent multi-agent

점수: 6/10 — 점수 6/10: multi-agent


[6/10] Deep Incentive Design with Differentiable Equilibrium Blocks Automated design of

Deep Incentive Design with Differentiable Equilibrium Blocks Automated design of multi-agent interactions with desirable equilibrium outcomes is inherently difficult due to the computational hardness, non-uniqueness, and instability of the resulting equilibria. In this work, we propose the use of game-agnostic differentiable equilibrium blocks (DEBs) as modules in a novel, differentiable framework to address a wide variety of incentive design problems from economics and computer science. We call this framework deep incentive design (DID). To validate our agent_orchestration agent multi-agent

점수: 6/10 — 점수 6/10: multi-agent


[6/10] Security Considerations for Artificial Intelligence Agents This article, a light

Security Considerations for Artificial Intelligence Agents This article, a lightly adapted version of Perplexity's response to NIST/CAISI Request for Information 2025-0035, details our observations and recommendations concerning the security of frontier AI agents. These insights are informed by Perplexity's experience operating general-purpose agentic systems used by millions of users and thousands of enterprises in both controlled and open-world environments. Agent architectures change core assumptions around code-data separation, authority boundaries, agent_orchestration ron agent multi-agent

점수: 6/10 — 점수 6/10: multi-agent


[6/10] Deep Incentive Design with Differentiable Equilibrium Blocks Automated design of

Deep Incentive Design with Differentiable Equilibrium Blocks Automated design of multi-agent interactions with desirable equilibrium outcomes is inherently difficult due to the computational hardness, non-uniqueness, and instability of the resulting equilibria. In this work, we propose the use of game-agnostic differentiable equilibrium blocks (DEBs) as modules in a novel, differentiable framework to address a wide variety of incentive design problems from economics and computer science. We call this framework deep incentive design (DID). To validate our agent_orchestration agent multi-agent

점수: 6/10 — 점수 6/10: multi-agent


[6/10] 코로나 제로 금리 시절 영끌로 산 다세대 아파트와 오피스 대출 물량의 약 60%가 2026년 하반기에 만기가 도래하는 대출 규모는 약 1조 달러

코로나 제로 금리 시절 영끌로 산 다세대 아파트와 오피스 대출 물량의 약 60%가 2026년 하반기에 만기가 도래하는 대출 규모는 약 1조 달러, 당시 3%대 변동 금리로 빌렸던 돈을 이제 7~8%가 넘는 금리로 차환해야 하는데 국채금리가 5% 수준이면 롤오버 불가능 = 건물주들이 건물을 포기하고 압류로 넘어가는 물량이 하반기에 쏟아질 것으로 예상

여기서 왜 미국 금리가 3.4~3.75%인데 저 글에 금리가 7~8%냐고 궁금한 사람들이 있을텐데 연준의 금리는 초단기 금리이고 기업이 상업용 부동산 대출을 할때는 장기 국채금리 (10년물, 30년물)을 기준으로 삼음 이 장기 국채금리가 데드라인 5%를 넘어서고 있다는 말임 아니 그럼 5%잖음? 왜 또 7~8%냐고 할텐데 돈을 때일 리스크에 대한 위험 가산금리(스프레드)가 붙음 그게 2~3%임

장기 국채금리+가산금리= 실제 대출 금리7~8%

최근 기관들 분석에 의하면 프라이빗 크레딧을 이용하는 기업들 약 40%가 마이너스 현금흐름 상태, 얘네들 이자도 못내서 이자를 빚에 얹어 나중에 갚는 현물이자 방식으로 연명중 = 얘네들도 2026년 하반기에 만기가 집중

CMBS라고 상업용 모기지 담보부 증권 절반 이상이 상환 불능 통보 글로벌 4대 신용평가사인 모닝스타 DBRS에서 내놓은 2026년 전망에 따르면 올해 만기로 돌아오는 1000억 달러 규모의 CMBS 대출중에 절반 이상이 원금 상환도 못하고 디폴트 될 거라는 분석을 공식적으로 발표함

그냥 내 머리로 내놓은 상황의 예측으로는 올해 중반까지 중소형 은행들과 프라이빗 크레딧 펀드사들이 장부에 부실을 숨기고 버티기로 들어갔고 우량 자산들 매각하면서 현금 쟁기기에 들어가 시장 유동성이 마름 (경기도웨일 칼럼 참고)

하반기부터 부동산 대출 만기 연장이 줄줄이 거부되고 부동산들 무더기 압류 사태 들어감

부실을 떠안은 프라이빗 크레딧 펀드사들이 고객들 투자금 환매 중단 선언하고 시장에서는 시스템 위기를 공식적으로 인정하고 패닉셀을 하지 않을까 함

이거 막을려면 제로금리 수준으로 돌아가야 하는데 (이 방법밖에 없음) 트럼프도 이 상황을 보고 받고 “파월 개새끼야 금리 좀 내려 병시나 너 공사비 횡령 고소” 이러지 않았을까 함

11월 3일 미국 중간선거에 공화당이 자빠질 위기인데 5월에 케빈 워시가 취임하고 6-7-9-10월 4번의 금리 결정에 어떤 스탠스를 취해서 위기를 넘길지 아니면 그냥 개박살 내고 빌미로 금리인하+양적완화를 할지 그건 계속 모니터링 해야 할 상황

뭐 그냥 개인적인 뷰니까 본인의 분석과 안맞다면 본인 스탠스대로 가도 됩니다. 이게 꼭 맞을거라는 보장은 없는거니까..

다들 굿밤 되시고 즐거운 주말 보내시길

점수: 6/10 — 점수 6/10: 분석